JP7600057B2 - Water treatment status monitoring system and water treatment status monitoring method - Google Patents

Description

The present invention relates to a water treatment status monitoring system and a water treatment status monitoring method.

Wastewater treatment systems in water supply and sewerage plants and other plants achieve the target water quality of the treated water by using various water treatment technologies according to the quality of the water being treated (also called "raw water"). Among the water treatment technologies, coagulation treatment, which removes solids from the water (turbidity (also called "turbidity")), requires appropriate chemical injection management. However, when the quality of the water being treated fluctuates, the amount of chemical injection required may depend on the experience of a skilled operator (expert).

Based on their experience, an expert can determine the appropriate amount of chemical injection based on the results of jar tests to check the flocculation state of turbidity, monitoring and control data, on-site visual inspection of the flocculation state of turbidity, and confirmation in the monitoring room of images of flocs formed by flocculation of turbidity. However, the flocculation state of turbidity observed visually by an expert is not recorded in the monitoring and control system (monitoring and control data). For this reason, the experience of an expert becomes tacit knowledge, and there is an issue that the knowledge necessary for chemical injection management is difficult to pass on to others who are not experts.

In the future, there will need to be a function that allows even non-experts to determine the flocculation state of turbidity and to determine early failure of flocculation. Therefore, progress is being made in the development of technology that uses sensing technology to obtain image data of flocs and uses machine learning and artificial intelligence to determine whether turbidity has flocculated properly.

Patent Document 1 discloses an information processing device (hereinafter, sometimes referred to as the "conventional device") described below. The conventional device performs deep learning on training data that associates a category indicating whether flocs formed in the future will have an adverse effect on the water treatment with each of 400 seconds of image data from when a known amount of coagulant is added to raw water to be treated and stirred.

Then, when image data (image data to be discriminated) is input, the conventional device discriminates, based on the deep learning results and the discrimination target image data, whether flocs that will be formed in the raw water corresponding to the discrimination target image data in the future will have an adverse effect on the drinking water treatment. The discrimination target image data is image data taken within 300 seconds after a known amount of coagulant is added to the raw water to be treated and stirred, and is not associated with a category indicating whether flocs that will be formed in the future will have an adverse effect on the drinking water treatment.

JP 2020-134284 A

In facilities that treat and purify treated water to obtain treated water of a target quality, there is a demand for technology that allows even unskilled personnel to easily determine whether the flocculation state of turbidity in the treated water shown in an image is good or bad. The present invention has been made to solve the above problem. In other words, one of the objects of the present invention is to provide a water treatment status monitoring system and a water treatment status monitoring method that can calculate information that allows even unskilled personnel to easily determine whether the flocculation state of turbidity in the treated water shown in an image is good or bad.

In order to solve the above problems, the water treatment status monitoring system of the present invention is applied to a facility that produces purified treated water by transporting the water to be treated to a plurality of treatment locations in sequence and performing a process including a coagulation and sedimentation process to remove turbidity contained in the water to be treated. The water treatment status monitoring system of the present invention includes a plurality of imaging devices that are installed at a plurality of different locations that can image the water to be treated and that image the water to be treated to obtain image data of the water to be treated for each location, including images of the water to be treated captured from each location; a treated water measuring device that measures water quality parameters related to the water to be treated, including the turbidity of the water to be treated, and obtains treated water data including the water quality parameters of the water to be treated; a treated water measuring device that measures the turbidity of the treated water and obtains treated water data including the turbidity of the treated water; and an information processing device that obtains the image data, the treated water data, and the treated water data from the imaging devices, the treated water measuring devices, and the treated water measuring devices. The information processing device is configured to use learning input data including the image data of the water to be treated and the treated water data as learning data, and to create, by machine learning, a plurality of coagulation quality judgment models that output information indicating the quality of the coagulation state of the turbidity of the water to be treated from input data including the image of the water to be treated, and the information processing device is configured to select at least one of the plurality of coagulation quality judgment models, acquire the image of the water to be treated from at least one of the plurality of imaging devices as an image for judgment, and use the selected coagulation quality judgment model to calculate information indicating the quality of the coagulation state of the turbidity of the water to be treated shown in the image for judgment from the input data for judgment including the image for judgment.

The water treatment status monitoring system of the present invention is applied to a facility that produces purified treated water by transporting the treated water to a plurality of treatment locations in sequence and performing a process including a coagulation and sedimentation process to remove turbidity contained in the treated water. The water treatment status monitoring system of the present invention includes a plurality of water sampling devices that sample the treated water present at a plurality of different locations, a plurality of imaging devices that capture images of the treated water present at the plurality of locations sampled by the plurality of water sampling devices to obtain sampled image data of the treated water for each location, including sampled images that are images of the treated water present at each location, a treated water measuring device that measures treated water quality parameters including treated water turbidity related to the treated water and obtains treated water data including the treated water quality parameters, a treated water measuring device that measures the treated water turbidity of the treated water and obtains treated water data including the treated water turbidity, and an information processing device that obtains the sampled water image data, the treated water data, and the treated water data from the imaging devices, the treated water measuring devices, and the treated water measuring devices. The information processing device is configured to use the learning input data including the sampled image data of the water to be treated and the treated water data as learning data, and to create, for each of the sampled image data of the water to be treated at each location, a coagulation quality judgment model that outputs information indicating the quality of the coagulation state of the turbidity of the water to be treated from the input data including the sampled image of the water to be treated by machine learning. The information processing device is configured to select at least one of the multiple coagulation quality judgment models created for each of the sampled image data of the water to be treated corresponding to each location, obtain the sampled image of the water to be treated from the imaging device as a judgment image, and use the selected coagulation quality judgment model to calculate information indicating the quality of the coagulation state of the turbidity of the water to be treated shown in the judgment image from the judgment input data including the judgment image.

The water treatment status monitoring method of the present invention is a water treatment status monitoring method applied to a facility that produces purified treated water by transporting the water to be treated to a plurality of treatment locations in sequence and performing a process including a coagulation and sedimentation process to remove turbidity contained in the water to be treated, and includes the steps of: capturing images of the water to be treated using a plurality of imaging devices installed at a plurality of different locations capable of capturing images of the water to be treated, thereby obtaining image data of the water to be treated for each location including images of the water to be treated captured from each location; measuring water quality parameters related to the water to be treated, including the turbidity of the water to be treated, using a water to be treated measuring device, and obtaining treated water data including the water quality parameters of the water to be treated; measuring the turbidity of the treated water using a treated water measuring device, and obtaining treated water data including the turbidity of the treated water; and , acquiring the image data, the treated water data, and the treated water data from the imaging device, the treated water measuring device, and the treated water measuring device, and the information processing device uses learning input data including the image data of the treated water and the treated water data as learning data, and creates multiple coagulation quality judgment models that output information indicating the quality of the coagulation state of the turbidity of the treated water from input data including the image of the treated water through machine learning, selects at least one of the multiple coagulation quality judgment models, acquires the image of the treated water from at least one of the multiple imaging devices as a judgment image, and uses the selected coagulation quality judgment model to calculate information indicating the quality of the coagulation state of the turbidity of the treated water shown in the judgment image from the judgment input data including the judgment image.

According to the present invention, even an unskilled person can calculate information that allows them to easily determine whether the flocculation state of the turbidity in the treated water shown in the image is good or bad.

FIG. 1 is a schematic diagram showing an example in which a water treatment status monitoring system according to a first embodiment of the present invention is applied to a water purification plant. FIG. 2 is a block diagram showing an example of the hardware configuration of an information processing device included in the water treatment status monitoring system. FIG. 3 is a table for explaining the test conditions of the test example and the treatment results of the water to be treated. FIG. 4 is a graph for explaining the change over time in the integrated value of the floc area. FIG. 5 is a flowchart showing a process flow executed by the model generating means. FIG. 6 is a flowchart showing a process flow executed by the model generating means. FIG. 7 is a flowchart showing a process flow executed by the coagulation quality determining means. FIG. 8 is a flow chart showing a process flow executed by the coagulation quality determining means. FIG. 9 is a flowchart showing a process flow executed by the model generating means of the first modified example. FIG. 10 is a flowchart showing a process flow executed by the model generating means of the first modified example. FIG. 11 is a graph for explaining an example of an evaluation result of the coagulation quality determination model. FIG. 12 is a flow chart showing a process flow executed by the coagulation quality determining means of the first modified example. FIG. 13 is a schematic diagram showing an example in which the water treatment status monitoring system according to the second embodiment of the present invention is applied to a water purification plant. FIG. 14 is a flowchart showing a process flow executed by the model generating means of the water treatment status monitoring system according to the third embodiment of the present invention. FIG. 15 is a flowchart showing a process flow executed by the model generating means of the water treatment status monitoring system according to the third embodiment. FIG. 16 is a flowchart showing a process flow executed by the model generating means of the water treatment status monitoring system according to the fourth embodiment of the present invention. FIG. 17 is a flowchart showing a process flow executed by the flocculation quality determining means of the water treatment status monitoring system according to the fourth embodiment.

<<Details of the background technology>>
First, in order to facilitate understanding of the present invention, details of the background art will be described. In some cases, the maintenance and management of plants such as water and sewage plants, chemical plants, power plants, and waste treatment plants is managed by relying on the visual inspection and experience of operators. When a plant is operating continuously for 24 hours, it is often the case that an operator is present at the plant at all times. However, since Japan's productive-age population (ages 15 to 64) is on the decline, it is necessary to start thinking now about sustainable solutions for when labor shortages occur in the future.

In recent years, with the improvement of computer computing power and communication infrastructure, it has become possible to easily handle large amounts of data. This has led to the evolution and spread of sensing technology (IoT technology) that replaces the five senses of operators. In addition, with the evolution of machine learning and artificial intelligence analysis algorithms (AI technology), there has been active consideration of the use of AI technology even in tasks that previously required human judgment and were difficult to mechanize or automate. For example, in order to continue plant maintenance work even in the event of a labor shortage, solutions are being offered that install cameras to replace the operator's vision, process captured images and videos, and diagnose changes in the appearance of equipment and products.

As described in [Background Art], wastewater treatment systems in water supply and sewage plants and other plants achieve the target water quality of the treated water by using various water treatment technologies according to the quality of the water being treated. Among water treatment technologies, coagulation treatment requires appropriate chemical injection management. However, the amount of chemical injection when the quality of the water being treated fluctuates may depend on the experience of a skilled operator. A skilled operator can determine the appropriate amount of chemical injection based on experience, from the results of jar tests to check the coagulation state of turbidity, monitoring and control data from the monitoring and control system that monitors the plant, on-site visual inspection of the coagulation state of turbidity, and confirmation of floc images in the monitoring room.

On the other hand, the aggregation state of turbidity observed visually by an expert is not recorded in the monitoring and control system (monitoring and control data). As a result, the experience of the expert becomes tacit knowledge, and there is an issue that the knowledge necessary for chemical injection management is difficult to pass on to people other than the expert.

In the future, in order to enable even unskilled personnel to determine the flocculation state of turbidity, monitoring and control systems (monitoring systems) will need to have the functionality to determine poor flocculation at an early stage. Therefore, technology is being developed that uses sensing technology to obtain image data of flocs and machine learning and artificial intelligence to determine whether the flocculation state of turbidity is good or bad (hereinafter referred to as "flocculation quality of turbidity" or simply "flocculation quality").

For example, the technology of Patent Document 1 mentioned above installs cameras in the rapid mixing basin and the flocculation basin, acquires image data (image data captured over a specified period of time) showing the formation of flocculation nuclei (changes over time) in the rapid mixing basin, and acquires image data (image data captured over a specified period of time) showing the initial state of floc formation in the flocculation basin, and determines whether flocs that will be formed in the future from the image data are good quality flocs based on the results of learning by deep learning using the acquired image data as teacher data.

However, while it is possible to capture images of the formation of agglomeration nuclei and flocs (changes over time) (to display the formation of agglomeration nuclei and flocs in image data for a specified period of time (image data for 400 seconds or 300 seconds)) using small test devices or batch-type devices, it is difficult to do so in an actual water supply and wastewater plant unless a camera is installed in an appropriate location. For this reason, the technology in Patent Document 1 may not be able to properly determine whether flocs are good or bad, depending on the specifications of the water supply and wastewater plant.

In contrast, one of the features of the present invention is to assume an actual continuous plant, and to obtain an image (hereinafter sometimes referred to as an "aggregation image") of the agglomeration state of turbid matter in the treated water taken at a certain time in a location where the agglomeration state of the turbid matter in the treated water can be imaged, where the difference between good and bad agglomeration is easy to determine, according to the quality of the treated water, and to use a model that uses the image as input and outputs information indicating the good or bad agglomeration state of the turbid matter in the treated water shown in the agglomeration image, and to calculate information indicating the good or bad agglomeration state from the acquired agglomeration image.

As a result, the present invention can calculate information that enables even an unskilled person to easily determine whether the aggregation state of the turbidity is good or bad from the aggregation image. Furthermore, the present invention allows even an unskilled person to easily determine whether the aggregation state of the turbidity is good or bad, and can therefore quickly determine whether the turbidity is poorly aggregated.

<<Embodiment>>
Hereinafter, embodiments of the present invention will be described with reference to the drawings.

<<First embodiment>>
<Configuration>
A water treatment status monitoring system according to a first embodiment of the present invention (hereinafter, may be referred to as a "first monitoring system") will be described. Fig. 1 is a schematic diagram showing an example in which the first monitoring system is applied to a water purification plant 100, which is an example of a plant facility.

The water purification plant 100 is a facility that purifies river water, dam water, and groundwater and sends the treated water to factories, general households, etc., and is composed of multiple facilities. In this example, the water purification plant 100 includes a receiving well 110, a mixing tank 120, a flocculation tank 130, a sedimentation tank 140, and a filtration tank 150, as shown in FIG. 1. The treated water Wa1 is transported from a river, dam, etc. to the receiving well 110, the mixing tank 120, the flocculation tank 130, the sedimentation tank 140, and the filtration tank 150 in that order, and is purified, disinfected, etc., to remove turbidity from the treated water Wa1 and sterilize (disinfect) the treated water Wa1. The treated water Wa1 after purification and disinfection is called "treated water Wa2". The treated water Wa2 is supplied to households, factories, etc. as tap water.

The first monitoring system is applied to the water purification plant 100 and includes a first water quality meter 210a and a second water quality meter 210b, a first imaging device 220a and a second imaging device 220b, and an information processing device 230. These are connected to each other so that they can communicate with each other by wire or wirelessly. In the following description, the "first water quality meter 210a" and the "second water quality meter 210b" are referred to as "water quality meter 210" when there is no particular need to distinguish them. The "first water quality meter 210a" may also be referred to as "treated water measuring device" for convenience. The "second water quality meter 210b" may also be referred to as "treated water measuring device" for convenience. The "first imaging device 220a" and the "second imaging device 220b" are referred to as "imaging device 220" when there is no particular need to distinguish them.

The water quality meter 210 is a sensor group including multiple sensors that measure water quality items related to water treatment. The multiple sensors are, for example, a turbidity meter that measures the turbidity (turbidity) of the water, a water thermometer that measures the water temperature, a pH meter that measures the pH, an alkalinity meter that measures the alkalinity, a TOC meter that measures organic matter in the water, an ultraviolet spectrometer, and a water flow meter that measures the amount of treatment, although not the water quality.

The first water quality meter 210a is installed at a position where it can measure the water Wa1 being treated at the water purification plant 100. In this example, the first water quality meter 210a is installed at a position where it can measure the water Wa1 being treated at the receiving well 110. The second water quality meter 210b is installed at a position where it can measure the treated water Wa2.

The first water quality meter 210a measures the water to be treated Wa1 at predetermined time intervals and transmits the measurement values (measurement values of each sensor) to the information processing device 230. For convenience, the measurement values of each sensor for the water to be treated Wa1 may also be referred to as "water quality parameters of the water to be treated." The second water quality meter 210b measures the water to be treated Wa2 at predetermined time intervals and transmits the measurement values (measurement values of each sensor) to the information processing device 230.

Each of the first imaging device 220a and the second imaging device 220b is installed at a position where it can image the flocculation state of the turbidity of the treated water Wa1 during the process of flocculation nucleation and floc growth after the injection of the chemical. Each of the first imaging device 220a and the second imaging device 220b may be an imaging device having a mechanism for recording videos or images at regular time intervals. In this example, each of the first imaging device 220a and the second imaging device 220b is a camera that captures videos. In this example, the first imaging device 220a is installed at a position where it can image the treated water Wa1 at the inlet of the flocculation tank 130. The second imaging device 220b is installed at a position where it can image the treated water Wa1 at the outlet of the flocculation tank 130. The installation positions of the first imaging device 220a and the second imaging device 220b are not limited to this, and for example, the first imaging device 220a may be installed in a position where it can image the treated water Wa1 in the mixing tank 120.

The imaging device 220 may image the water Wa1 to be treated while the imaging device 220 is directly immersed in the water, may image the water surface, or may image the water Wa1 to be treated through a transparent wall. In other words, the method by which the imaging device 220 images the water Wa1 to be treated is not particularly limited as long as it is capable of imaging (capable of imaging) the flocculated state of turbidity.

The imaging device 220 uses image extraction means (not shown) to extract images (still images) at predetermined time intervals and transmits them to the information processing device 230. It is desirable that the time interval for extracting images be matched (the same) as the measurement time interval of the water quality meter 210. The image extraction means may be provided in the imaging device 220 or in the information processing device 230. If the information processing device 230 is provided with image extraction means, the image extraction means processes the video received from the imaging device 220 as described above.

The information processing device 230 includes an information storage means 231, a model generation means 232, an aggregation pass/fail determination means 233, and an aggregation pass/fail output means 234.

The information storage means 231 adds time information of the time when the measurement values were measured to the measurement values (e.g., turbidity, water temperature, pH, alkalinity, etc.) transmitted from the first water quality meter 210a, and holds (memorizes, stores) the measurement values (multiple "sets of measurement values") as time series data (hereinafter referred to as "treated water data"). The information storage means 231 adds time information of the time when the measurement values were measured to the measurement values (e.g., turbidity, water temperature, pH, alkalinity, etc.) transmitted from the second water quality meter 210b, and holds (memorizes, stores) the measurement values (multiple "sets of measurement values") as time series data (hereinafter referred to as "treated water data").

The information storage means 231 adds time information of the time when the first image was captured to the image (hereinafter also referred to as "first image") transmitted from the first imaging device 220a at a predetermined time interval, and holds (stores) the multiple first images at the predetermined time interval as time series data (hereinafter referred to as "first image data"). The information storage means 231 adds time information of the time when the second image was captured to the image (hereinafter also referred to as "second image") transmitted from the second imaging device 220b at a predetermined time interval, and holds (stores) the multiple second images at the predetermined time interval as time series data (hereinafter referred to as "second image data"). The information storage means 231 may further add information indicating the location where the image was captured to each image.

The information storage means 231 associates and holds (memorizes, stores) the treated water data, the treated water data, the first image data, and the second image data based on time information. For example, the information storage means 231 combines (associates) the treated water data acquired by the first water quality meter 210a, the treated water data acquired by the second water quality meter 210b, and the image data acquired by the first imaging device 220a and the second imaging device 220b at the same time (same time range) using time as a key, and holds (memorizes, stores) these as a data set.

For example, each data may be associated so that it corresponds to the process of treating a certain treated water Wa1 when the treated water Wa1 is treated at the water purification plant 100. In this case, for example, the information storage means 231 preferably associates the measurement value of the treated water Wa1 at a certain time t1, the first image at time t1+ta, the second image at time t1+tb, and the measurement value of the treated water at time t1+tc with each other based on the time information. ta corresponds to the time from the installation position of the first water quality meter 210a until the treated water Wa1 is transported to the first imaging device 220a, tb corresponds to the time from the installation position of the first water quality meter 210a until the treated water Wa1 is transported from the installation position of the first water quality meter 210a until the installation position of the second imaging device 220b, and tc corresponds to the time from the installation position of the first water quality meter 210a until the treated water Wa1 is transported from the installation position of the first water quality meter 210a until the installation position of the second water quality meter 210b. In addition, because the water quality of the water to be treated Wa1 typically does not fluctuate over a short period of time, as described above, each piece of data (each image, each measurement value) can be associated at the same time, and in this case too, data corresponding to the treatment progress of the water to be treated Wa1 can be obtained.

The model generating means 232 uses the data stored in the information storing means 231 to generate (create) a model for determining whether the turbidity of the turbidity reflected in the image of the treated water Wa1 is coagulated or not (a model that outputs information indicating whether the coagulation state of the turbidity of the treated water Wa1 reflected in the image is good or not). More specifically, the model generating means 232 generates a first coagulation good or not judgment model using a data set including the treated water data, the treated water data, and the first image data, which are associated with each other. The model generating means 232 generates a second coagulation good or not judgment model using a data set including the treated water data, the treated water data, and the second image data, which are associated with each other. The model generating means 232 holds (stores) the generated first coagulation good or not judgment model and second coagulation good or not judgment model.

The coagulation quality determination means 233 uses either the first coagulation quality determination model or the second coagulation quality determination model generated by the model generation means 232 to calculate information indicating the quality of the coagulation state of the turbidity in the treated water Wa1 from input data including an image (first image or second image) showing the coagulation state of the turbidity in the treated water Wa1 at a certain time (for example, the current time). The coagulation quality determination means 233 is capable of selecting (setting) the model used to calculate the information indicating the quality of the coagulation state of the turbidity to either the first coagulation quality determination model or the second coagulation quality determination model based on the user's operation on an operation device not shown.

The coagulation pass/fail output means 234 outputs the information calculated by the coagulation pass/fail determination means 233.

Figure 2 is a schematic diagram showing an example of the hardware configuration of the information processing device 230. As shown in Figure 2, the information processing device 230 includes a CPU 241, a ROM 242, a RAM 243, a non-volatile storage device (HDD) 244 that can read and write data, a network interface 245, and an input/output interface 246. These are connected to each other via a bus 247 so that they can communicate with each other. Note that the information processing device 230 may be composed of multiple information processing devices.

The CPU 241 loads various programs (not shown) stored in the ROM 242 and/or HDD 244 into the RAM 243 and executes the programs loaded into the RAM 243 to realize various functions. As described above, various programs to be executed by the CPU 241 are loaded into the RAM 243, and data used when the CPU 241 executes the various programs is temporarily stored. The ROM 242 and/or HDD 244 are non-volatile storage media in which various programs are stored. The network interface 245 is an interface for connecting the information processing device 230 to a network. The input/output interface 246 is an interface for connecting to operating devices such as a keyboard and a mouse, and a display.

The information storage means 231 is composed of a HDD 244, a network interface 245 and/or an input/output interface 246. The model generation means 232 is composed of a program stored in a ROM 242 and/or a HDD 244 executed by a CPU 241 of the information processing device 230, and the HDD 244. The coagulation quality determination means 233 is composed of a program stored in a ROM 242 and/or a HDD 244 executed by a CPU 241 of the information processing device 230. The coagulation quality output means 234 is composed of a network interface 245 and/or an input/output interface 246.

The information processing device 230 is connected to a monitoring control device 310. The monitoring control device 310 is connected to a chemical pump 320. The monitoring control device 310 is configured to be able to control the amount of coagulant injected into the treated water Wa1 in the mixing tank 120 by controlling the chemical pump 320. The monitoring control device 310 and the chemical pump 320 may be included in the first monitoring system. The functions of the monitoring control device 310 may be included in the information processing device 230.

<Test Example>
In the water purification plant 100, the process of removing turbidity from the water to be treated Wa1 is called "coagulation and sedimentation treatment." In a typical water purification plant 100, the coagulation and sedimentation treatment involves injecting a chemical called a coagulant, such as polyaluminum chloride or aluminum sulfate, into the water to be treated Wa1 that has flowed into the facility. One of the purposes of injecting the chemical is to remove insoluble substances (turbidity) suspended in the water. The turbidity is negatively charged in the water, but adding the chemical makes it possible to electrically neutralize it. Another purpose of injecting the chemical is to facilitate the coagulation of the turbidity when neutralized turbidity collides with each other due to the bridging action of the chemical.

In the mixing tank 120 (rapid mixing tank), a chemical is injected into the water Wa1 to be treated, and rapid mixing causes the turbid matter to collide with each other, forming flocculation nuclei in the water Wa1 to be treated. Next, in the flocculation tank 130 (flocculation tank), flocs are grown by slow mixing at a strength that does not destroy them. Then, in the settling tank 140 (sedimentation tank), the grown flocs are settled and removed.

Here are some examples of factors that affect the floc coagulation mechanism. As the amount of chemical injected increases, the bridging action becomes stronger, which leads to faster floc growth, a better coagulation state of the turbidity, and a lower turbidity of the treated water (hereinafter also referred to as "treated water turbidity"). However, if too much chemical is injected, the turbidity becomes positively charged, resulting in poor coagulation of the turbidity and a possible increase in the turbidity of the treated water.

Increasing the mixing intensity increases the frequency of collisions between turbid matter, and so the higher the mixing intensity, the faster the flocs will grow. However, if the mixing intensity is too strong, it can cause the grown flocs to be destroyed.

The more the amount of turbidity in the water (the turbidity of the water to be treated (hereinafter also referred to as "turbidity of the water to be treated")) increases, the more frequently the suspended solids (or flocs) collide with each other, and therefore the faster the flocs grow.

As described above, floc growth is affected by plant operating conditions, such as the amount of chemicals injected and mixing strength, and water quality conditions.

If an experienced worker can visually inspect the site or look at the screen in the monitoring room showing the flocculation state of the turbidity and, based on experience, confirm that the flocs are smaller than normal, he or she will determine that flocculation is poor and take appropriate measures to purify the water.

However, as mentioned above, Japan's working-age population (15-64 years old) is on a downward trend, so we need to start thinking now about sustainable solutions for when labor shortages occur in the future.

In response to this, the information processing device 230 of the first monitoring system generates a model capable of calculating information indicating whether the flocculation state of the turbidity of the treated water Wa1 is good or bad from the flocculation image so that even an unskilled person can judge whether the flocculation state of the flocs (turbidity) shown in the image (screen) is good or bad. That is, the information processing device 230 of the first monitoring system uses data including an image at a certain time as input data, detects changes in the flocculation state (differences in the flocculation state appearing in the image at a certain time), and generates a model capable of calculating information indicating whether the flocculation state of the turbidity of the treated water Wa1 is good or bad from the input data. Then, the first monitoring system uses the model to calculate information indicating whether the flocculation state of the turbidity of the treated water Wa1 is good or bad from the flocculation image (in this example, a predicted value of the treated water turbidity (sometimes referred to as "predicted treated water turbidity"). Note that the smaller the predicted treated water turbidity, the better the flocculation state of the treated water Wa1 is (good), and the larger the predicted treated water turbidity, the worse the flocculation state of the treated water Wa1 is (bad). In order to confirm whether the change in the flocculation state, which has a correlation with the turbidity of the treated water, can be detected by an image, a flocculation image acquisition test was performed. FIG. 3 shows the test conditions for the flocculation image acquisition test and the treatment result of the treated water (turbidity of the treated water). In the flocculation image acquisition test, a blackboard (background board) that serves as the background when the flocculation state is imaged by the agitator and the camera is installed at a position 15 mm from the inner wall of the tank in a rectangular transparent tank, and the camera is installed on the outer wall side of the tank so as to image the background board. In addition, the light source is located at the top of the tank. The treated water was adjusted by adding a kaolin reagent as a simulated turbidity to tap water. The water quality of each test condition at this time is as shown in FIG. 3, and the flocculation conditions were changed by changing the amount of the agent injected into the treated water adjusted to turbidity of 25 degrees, 81 degrees, and 85 degrees. After starting imaging with the camera, a preset amount of agent injection flocculant was added, and rapid stirring was performed for 3 minutes, slow stirring for 10 minutes, and standing for 10 minutes. The supernatant water of the tank was then collected and the turbidity of the treated water was measured. The rapid mixing process, which corresponds to the rapid mixing basin (mixing tank 120), had a mixing intensity of 148.2 ^s-1 , and the slow mixing process in the flocculation basin (flocculation tank 130) had a mixing intensity of 64.0 s ^-1 . After the 10-minute stand was completed, the camera stopped taking images. During the test, the camera captured images of the flocs passing between the inner wall of the tank and the background plate.

Images were cut out from the captured video and image processing was performed to calculate the area value (in pixels) of the flocs in the image. In this test example, agglomeration images were cut out from the captured video in 0.5 second intervals. Image processing included grayscaling, binarization and object recognition (floc detection). Since calculations from only one image may result in inconsistent results, the floc area value was calculated from agglomeration images cut out in 0.5 second intervals from the start of slow stirring (0 minutes), and then integrated over 10 seconds to obtain the integrated floc area value. The calculation interval for the integrated floc area value was 30 seconds (0 minutes, 0.5 minutes, 1 minute, ...).

Figure 4 shows the change in the floc area integrated value over time. The floc area integrated value increased as the treatment time progressed. This is due to the growth of flocs. Runs 1 and 2 had smaller floc area integrated values than Runs 3 and 4. This is because the turbidity of the treated water was approximately 25 degrees, which is lower than the approximately 80 degrees of Runs 3 and 4, and therefore less flocs were generated. Also, the increase in the floc area integrated value was more gradual in Runs 1 and 2 than in Runs 3 and 4.

This is due to the floc coagulation mechanism mentioned above, and the higher the turbidity of the treated water, the more frequent the collisions of turbid matter (or coagulation nuclei, flocs) can be, leading to faster floc growth. From the above, in Runs 1 and 2, where the turbidity of the treated water is lower than in Runs 3 and 4, the growth of flocs is relatively slower and the increase in the floc area value is more gradual. Furthermore, in Runs 1 and 2, the higher the amount of coagulant injected, the earlier the increase in the floc area integrated value begins. This is because there is also a bridging effect between the coagulants, which increases the probability of coagulation when turbid matter (or coagulation nuclei, flocs) collide.

Comparing Run 2 and Run 4, the final treated water turbidity is about the same, but the trends and maximum values of the floc area integrated value are different.

For these reasons, when judging the quality of the flocculation state of suspended solids from images, it is necessary to take into account the water quality, particularly the turbidity of the water being treated.

Specifically, when judging the quality of coagulation from images of the treated water in Run 1 and Run 2 (calculating information showing the quality of the coagulation state (predicted treated water turbidity)), it is best to use images from the treatment time at which a difference in the integrated floc area value occurs. In the case of Run 1 and Run 2, it is considered that 1 to 7 minutes, especially 3 to 5 minutes, is good. In addition, in the case of Run 3 and Run 4, because a difference appears immediately after the start of slow stirring, it is possible to judge the quality of coagulation (calculate information showing the quality of the coagulation state (predicted treated water turbidity)) even if an image of the treated water at 0 minutes is used.

From the above, it can be seen that a change in the aggregation state of turbidity correlated with the turbidity of the treated water (a difference in the aggregation state of turbidity that appears in the image) can be detected from an image taken at a specific time (hour). Furthermore, it can be seen that the formation behavior of the aggregation of turbidity (floc formation behavior) that appears in the image differs depending on the water quality of the treated water (particularly the turbidity of the treated water). It can be seen that the elapsed time at which the difference in the aggregation state of turbidity that appears in the aggregation image correlated with the turbidity of the treated water appears large (or small) varies depending on the water quality of the treated water. Therefore, if an image taken at an appropriate specific elapsed time (hour) based on the water quality of the treated water (i.e., an image taken at a place corresponding to the elapsed time of treatment appropriate for obtaining an image of the treated water) is used as input data, the difference in the aggregation state of turbidity that appears in the image correlated with the turbidity of the treated water will appear large, and it can be seen that the quality of the aggregation state of the turbidity of the treated water can be determined more accurately (information indicating the quality of the aggregation state can be calculated). Furthermore, if parameters indicating the water quality of the water being treated, which are correlated with the turbidity of the treated water, are also used as input data, it is possible to more accurately determine whether the flocculation state of the turbidity in the water being treated is good or bad (to calculate information indicating the quality of the flocculation state).

<Outline of operation>
As described above, the information processing device 230 generates the first coagulation quality judgment model and the second coagulation quality judgment model. When calculating information indicating the quality of the coagulation state of the turbidity of the treated water Wa1 (in this example, the predicted value of the treated water turbidity (predicted treated water turbidity) of the treated water Wa2), the information processing device 230 selects one of the first coagulation quality judgment model and the second coagulation quality judgment model as a model to be used for calculating the information indicating the quality of the coagulation state of the turbidity. For example, an operation device (not shown) is connected to the information processing device 230, and based on the operation of the operation device by the user, one of the first coagulation quality judgment model and the second coagulation quality judgment model is selected as a model to be used for calculating the information indicating the quality of the coagulation state of the turbidity.

The information processing device 230 calculates information indicating whether the flocculation state of the turbidity in the treated water Wa1 is good or bad using one of the selected first flocculation good/bad judgment model and second flocculation good/bad judgment model. For example, the user appropriately selects the optimal flocculation good/bad judgment model via an operation device (not shown) according to the turbidity of the treated water Wa1 measured in advance, and the information processing device 230 calculates information indicating whether the flocculation state of the turbidity in the treated water Wa1 is good or bad using the optimal flocculation good/bad judgment model according to the water quality (turbidity of the treated water) of the treated water Wa1. As a result, the information processing device 230 can calculate information that can easily determine whether the flocculation state of the turbidity in the treated water Wa1 is good or bad using the optimal flocculation good/bad judgment model as appropriate.

<Specific operation>
Fig. 5 is a flowchart showing the process flow executed by the model generating means 232. The model generating means 232 executes the flowchart shown in Fig. 5. Therefore, the model generating means 232 starts the process from step 500 in Fig. 5, and executes the processes from step 505 to step 520 described below in order. After that, the model generating means 232 proceeds to step 595 and temporarily ends this process flow.

Step 505: The model generating means 232 acquires a data set of the first image data, the treated water data, and the treated water data, which are associated with each other, from the information storing means 231. In this example, the treated water data includes, for example, the treated water turbidity, water temperature, pH, alkalinity, TOC (Total Organic Carbon), and coagulant injection concentration.

Step 510: The model generating means 232 performs image processing on each first image of the first image data. Specifically, the model generating means 232 grayscales each first image of the first image data and then performs brightness correction.

Step 515: The model generation means 232 normalizes the treated water data (each measured value of the treated water data).

Step 520: The model generation means 232 uses the processed first image data, the processed treated water data, and the treated water data as learning data, and generates a trained model (first coagulation pass/fail judgment model) by deep learning, using the first image (first image after image processing) and each measured value of the treated water data (each measured value after normalization processing) as input, and outputting predicted treated water turbidity. The model generation means 232 retains (stores) the generated first coagulation pass/fail judgment model.

In deep learning, for example, the first image after image processing and each measurement value of the treated water data (each measurement value after normalization processing), which are associated with each other using time as a key, are used as input, and one learning is performed using the treated water turbidity of the treated water data associated with the first image and each measurement value as the correct value. This learning is performed for each of the first images, each measurement value of the treated water data, and the treated water turbidity of the treated water data, which are associated based on the time of the learning dataset. In other words, multiple iterative learning is performed.

In deep learning, for a first image input, a known CNN is used to output it to a first output neuron before the final output, and for each measurement value input, a known NN is used to output it to a second output neuron before the final output. In the final stage, the outputs from the first and second output neurons are input to a connection layer, and then a final predicted value (predicted treated water turbidity) is output by a function. The error between the predicted value and the correct value is evaluated by the function, and learning is performed so that the function is minimized.

Figure 6 is a flowchart showing the processing flow executed by the model generation means 232. The model generation means 232 executes the flowchart shown in Figure 6. Therefore, the model generation means 232 starts processing from step 600 in Figure 6, and executes the processing from step 605 to step 620 described below in order. After that, the model generation means 232 proceeds to step 695 and temporarily ends this processing flow.

Step 605: The model generation means 232 acquires data sets of second image data, treated water data, and treated water data that are associated with each other from the information storage means 231.

Step 610: The model generating means 232 performs image processing on each second image of the second image data. Specifically, the model generating means 232 grayscales each second image of the second image data and then performs brightness correction.

Step 615: The model generation means 232 normalizes the treated water data (each measured value of the treated water data).

Step 620: The model generation means 232 uses the processed second image data, the processed treated water data, and the treated water data as learning data, and generates a trained model (second flocculation pass/fail judgment model) by deep learning similar to that described above, using the second image (second image after image processing) and each measurement value of the treated water data (each measurement value after normalization processing) as inputs, and outputting predicted treated water turbidity. The model generation means 232 retains (stores) the generated second flocculation pass/fail judgment model.

Figure 7 is a flowchart showing the processing flow executed by the coagulation quality determination means 233. This processing flow is executed by the coagulation quality determination means 233 when the first coagulation quality determination model is selected as the model used to calculate information indicating the quality of the coagulation state of the turbid matter. When the first coagulation quality determination model is selected, the coagulation quality determination means 233 executes the flowchart shown in Figure 7. Therefore, the coagulation quality determination means 233 starts processing from step 700 in Figure 7, and sequentially executes the processing of steps 705 to 720 described below. Thereafter, the coagulation quality determination means 233 proceeds to step 795 and temporarily ends this processing flow.

Step 705: The flocculation quality determination means 233 acquires a first image at the current time from the first imaging device 220a, and acquires treated water data (measurement values at the current time) from the first water quality meter 210a.

Step 710: The agglomeration quality determination means 233 processes the first image. Specifically, the agglomeration quality determination means 233 converts the first image into a grayscale image and then performs brightness correction.

Step 715: The coagulation pass/fail determination means 233 normalizes the treated water data (each measurement value at the current time).

Step 720: The coagulation quality judgment means 233 acquires the first coagulation quality judgment model from the model generation means 232, inputs the first image that has been image-processed and the treated water data that has been normalized into the first coagulation quality judgment model, and calculates and acquires a predicted value of the treated water turbidity (predicted treated water turbidity), which is the output of the first coagulation quality judgment model.

Figure 8 is a flowchart showing the processing flow executed by the coagulation quality judgment means 233. This processing flow is executed by the coagulation quality judgment means 233 when the second coagulation quality judgment model is selected as the model to be used for calculating information indicating the quality of the coagulation state of the turbid matter. Note that this processing flow is the same as that of Figure 7 except that the first image is replaced by the second image and the first coagulation quality judgment model is replaced by the second coagulation quality judgment model in Figure 7, so a detailed description will be omitted.

＜Effects＞
As described above, the first monitoring system can calculate information (predicted treated water turbidity) indicating the quality of the flocculation state of the turbidity in the treated water Wa1 shown in the image by appropriately using an optimal flocculation quality judgment model. In the first monitoring system, knowledge about the flocculation state of the turbidity, which was previously implicit knowledge of experts and relies on vision, is made explicit, so that even an inexperienced operator (unskilled person) can easily and quickly judge the quality of the flocculation state of the turbidity shown in the image by using this first monitoring system. Furthermore, in the past, when the water quality fluctuated, the operation was often performed on the safe side by increasing the amount of flocculant injection, but the first monitoring system can quickly judge the quality of the flocculation state of the turbidity in the treated water Wa1 shown in the image and operate with the optimal amount of flocculant injection, thereby reducing the amount of flocculant injection and reducing the cost of excess flocculant, thereby enabling cost savings.

<<First Modification>>
A first modified example of the first monitoring system will be described. In the first modified example, a first flocculation quality judgment model and a second flocculation quality judgment model are generated using a part of the data stored in the information storage means 231. In the first modified example, a threshold turbidity is obtained for judging whether the first flocculation quality judgment model or the second flocculation quality judgment model is used to reduce the error of the predicted value of the turbidity of the treated water (predicted treated water turbidity) by using at least a part (all of the data in this example) of the remaining data stored in the information storage means 231. In the first modified example, the turbidity of the treated water Wa1 is compared with the threshold turbidity, and one of the first flocculation quality judgment model and the second flocculation quality judgment model is selected based on the comparison result. In the first modified example, the predicted treated water turbidity of the treated water Wa2 is calculated using the selected model. Other than the above points, the first modified example is the same as the first monitoring system.

The following will focus on these differences.

<Specific operation>
Fig. 9 is a flowchart showing a process flow executed by the model generating means 232 of the first modified example. The model generating means 232 executes the flowchart shown in Fig. 9. Therefore, the model generating means 232 starts the process from step 900 in Fig. 9, and executes the processes of steps 905 to 935 described below in order.

Step 905: The model generation means 232 acquires data sets of the first image data, the second image data, the treated water data, and the treated water data, which are associated with each other, from the information storage means 231. At this time, the model generation means 232 acquires a data set that is a portion (70% in this example) of all data sets stored in the information storage means 231. The model generation means 232 uses the acquired data sets to execute the processing of steps 910 to 930 described below.

Step 910: The model generating means 232 performs image processing on each first image of the first image data. Specifically, the model generating means 232 grayscales each first image of the first image data and then performs brightness correction.

Step 915: The model generation means 232 normalizes the treated water data (each measured value of the treated water data).

Step 920: The model generation means 232 uses the processed first image data, the processed treated water data, and the treated water data as learning data, and generates a trained model (first coagulation pass/fail judgment model) by deep learning, which uses the first image (first image after image processing) and each measured value of the treated water data (each measured value after normalization processing) as input, and outputs predicted treated water turbidity. The model generation means 232 holds (stores) the generated first coagulation pass/fail judgment model.

Step 925: The model generating means 232 performs image processing on each second image of the second image data. Specifically, the model generating means 232 grayscales each second image of the second image data and then performs brightness correction.

Step 930: The model generation means 232 uses the processed second image data, the processed treated water data, and the treated water data as learning data, and generates a trained model (second flocculation pass/fail judgment model) by deep learning, which uses the second image (second image after image processing) and each measurement value of the treated water data (each measurement value after normalization processing) as input, and outputs predicted treated water turbidity. The model generation means 232 holds (stores) the generated second flocculation pass/fail judgment model.

Step 935: The model generation means 232 executes the evaluation process shown in the flowchart in FIG. 10.

Figure 10 is a flowchart showing the processing flow executed by the model generation means 232 of the first modified example. When the model generation means 232 proceeds to step 935 in Figure 9, it starts processing from step 1000 in Figure 10 and sequentially executes the processing from step 1005 to step 1035 described below.

Step 1005: The model generation means 232 acquires data sets of the first image data, the second image data, the treated water data, and the treated water data, which are associated with each other, from the information storage means 231. At this time, the model generation means 232 acquires all of the remaining data sets (30% in this example) of all data sets stored in the information storage means 231. The model generation means 232 uses the acquired data sets to sequentially execute the processes of steps 1010 to 1035 described below.

Step 1010: The model generating means 232 performs image processing on each first image of the first image data. Specifically, the model generating means 232 grayscales each first image of the first image data and then performs brightness correction.

Step 1015: The model generation means 232 normalizes the treated water data (each measured value of the treated water data).

Step 1020: The model generating means 232 calculates a predicted value of treated water turbidity (predicted treated water turbidity) using the first image data after image processing, the treated water data after normalization processing, and the first coagulation pass/fail judgment model. The model generating means 232 calculates the error of the predicted value (= | predicted treated water turbidity - treated water turbidity of treated water data (actual value) |) using the predicted value of treated water turbidity (predicted treated water turbidity) and the treated water turbidity of the treated water data.

Step 1025: The model generating means 232 performs image processing on the second image data. Specifically, the model generating means 232 grayscales each second image of the second image data and then performs brightness correction.

Step 1030: The model generating means 232 calculates a predicted value of treated water turbidity (predicted treated water turbidity) using the second image data after image processing, the treated water data after normalization processing, and the second coagulation pass/fail judgment model. The model generating means 232 calculates the error of the predicted value (= | predicted treated water turbidity - treated water turbidity of treated water data (actual value) |) using the predicted value of treated water turbidity (predicted treated water turbidity) and the treated water turbidity of the treated water data.

Step 1035: The model generating means 232 calculates a threshold turbidity based on the treated water turbidity of the treated water data, the error of the predicted value calculated in step 1020, and the error of the predicted value calculated in step 1030. For example, the threshold turbidity is the turbidity at the intersection P10 of the lines a1 and a2 when the graph shown in FIG. 11 is created. The line a1 is a line plotted based on the treated water turbidity of the treated water data and the error of the predicted value calculated in step 1020, and shows the relationship between the treated water turbidity of the treated water data and the average error of the predicted value when the first coagulation good/bad model is used. The line a2 is a line plotted based on the treated water turbidity of the treated water data and the error of the predicted value calculated in step 1030, and shows the relationship between the treated water turbidity of the treated water data and the average error of the predicted value when the second coagulation good/bad model is used.

According to the graph in FIG. 11, when the turbidity of the treated water Wa1 is greater than the threshold turbidity, the average error is smaller when the first coagulation quality judgment model is used than when the second coagulation quality judgment model is used, and therefore the accuracy of the predicted value (predicted treated water turbidity) of the treated water turbidity is higher. According to the graph in FIG. 11, when the turbidity of the treated water Wa1 is less than the threshold turbidity, the average error is smaller when the second coagulation quality judgment model is used than when the first coagulation quality judgment model is used, and therefore the accuracy of the predicted value (predicted treated water turbidity) is higher.

Then, the model generation means 232 proceeds to step 1095 to temporarily end this processing flow, and then proceeds to step 995 to temporarily end the processing flow of FIG. 9.

Figure 12 is a flowchart showing the process flow executed by the flocculation quality determination means 233 of the first modified example. The flocculation quality determination means 233 executes the flowchart shown in Figure 12. Therefore, the flocculation quality determination means 233 starts processing from step 1200 in Figure 12 and proceeds to step 1205, where it acquires the treated water data at the current time (each measured value at the current time) from the first water quality meter 210a.

Then, the flocculation pass/fail determination means 233 proceeds to step 1210 and determines whether the treated water turbidity of the treated water data is greater than the threshold turbidity.

If the turbidity of the treated water is greater than the threshold turbidity, the flocculation pass/fail judgment means 233 judges "Yes" in step 1210 and executes the processes of steps 1215 to 1230 described below in order, and then proceeds to step 1295 to temporarily end this processing flow.

Step 1215: The aggregation quality determination means 233 acquires the first image at the current time from the first imaging device 220a.

Step 1220: The agglomeration quality determination means 233 processes the first image. Specifically, the agglomeration quality determination means 233 converts the first image into a grayscale image and then performs brightness correction.

Step 1225: The coagulation pass/fail determination means 233 normalizes the treated water data (each measured value at the current time).

Step 1230: The coagulation quality judgment means 233 acquires the first coagulation quality judgment model from the model generation means 232, inputs the first image that has been image-processed and the treated water data that has been normalized into the first coagulation quality judgment model, and calculates and acquires a predicted value of the treated water turbidity (predicted treated water turbidity), which is the output of the first coagulation quality judgment model.

If the turbidity of the treated water is equal to or lower than the threshold turbidity, the flocculation pass/fail judgment means 233 judges "No" in step 1210 and executes the processes of steps 1235 to 1250 described below in order, and then proceeds to step 1295 to temporarily end this processing flow.

Step 1235: The aggregation quality determination means 233 acquires the second image at the current time from the second imaging device 220b.

Step 1240: The aggregation quality determination means 233 processes the second image. Specifically, the aggregation quality determination means 233 converts the second image into a grayscale image and then performs brightness correction.

Step 1245: The coagulation pass/fail determination means 233 normalizes the treated water data (each measurement value at the current time).

Step 1250: The coagulation quality judgment means 233 acquires the second coagulation quality judgment model from the model generation means 232, inputs the image-processed second image and the normalized treated water data into the second coagulation quality judgment model, and calculates and acquires a predicted value of treated water turbidity (predicted treated water turbidity), which is the output of the second coagulation quality judgment model.

＜Effects＞
The first modified example selects an optimal coagulation quality judgment model according to the turbidity of the treated water data, and uses the selected optimal coagulation quality judgment model to calculate information (predicted treated water turbidity) indicating the quality of the coagulation state of turbidity with higher accuracy. Note that the features of the first modified example may also be applied to the second and third embodiments described below.

<<Second embodiment>>
A water treatment status monitoring system according to a second embodiment of the present invention (hereinafter, may be referred to as a "second monitoring system") will be described.

Figure 13 is a schematic diagram showing an example of application of the second monitoring system to a water purification plant 100. As shown in Figure 13, the second monitoring system includes a first water sampling pump 1310a, a second water sampling pump 1310b, a first valve 1320a, a second valve 1320b, a flow cell 1330, piping 1340, and a valve control device 1350. In the second monitoring system, the second imaging device 220b of the first monitoring system is omitted, and the first imaging device 220a is installed in a position where it can image the flow cell 1330. Other than the above, the configuration is the same as that of the first monitoring system shown in Figure 1.

The first water sampling pump 1310a is installed at the inlet of the flocculation tank 130 and samples the water Wa1 to be treated at the inlet of the flocculation tank 130. The sampled water Wa1 to be treated flows into the flow cell 1330 through the pipe 1340. The first water sampling pump 1110a may also be installed at the mixing tank 120 and sample the water Wa1 to be treated from the mixing tank 120.

The second water sampling pump 1310b is installed at the outlet of the flocculation tank 130 and samples the water to be treated Wa1 from the outlet of the flocculation tank 130. The sampled water to be treated Wa1 passes through the pipe 1340 and flows into the flow cell 1330.

The flow cell 1330 is a transparent container through which the collected treated water Wa1 passes.

The first valve 1320a is an electromagnetic valve and is installed in the pipe 1340 between the first water collection pump 1310a and the branch point Pt1 of the pipe 1340. The first valve 1320a is set to either an open state in which the treated water Wa1 flows, or a closed state in which the flow of the treated water Wa1 is blocked, by controlling the flow of electricity through the first valve 1320a.

The second valve 1320b is an electromagnetic valve and is installed in the pipe 1340 between the second water collection pump 1310b and the branch point Pt1 of the pipe 1340. The second valve 1320b is set to either an open state in which the treated water Wa1 flows, or a closed state in which the flow of the treated water Wa1 is blocked, by controlling the supply of electricity to the second valve 1320b.

The valve control device 1350 controls the first valve 1320a to be set to either an open state or a closed state, and controls the second valve 1320b to be set to either an open state or a closed state. The valve control device 1350 controls the open/close states of the first valve 1320a and the second valve 1320b, thereby controlling the treated water Wa1 flowing into the flow cell 1330 so that the treated water Wa1 collected by the first water collection pump 1310a and the treated water Wa1 collected by the second water collection pump 1310b alternately flow into the flow cell 1330 every time a predetermined time has elapsed. That is, the valve control device 1350 controls the treated water Wa1 flowing into the flow cell 1330 so that the treated water Wa1 collected by the first water collection pump 1310a flows into the flow cell 1330 for a predetermined time, and the treated water Wa1 collected by the second water collection pump 1310b flows into the flow cell 1330 for a predetermined time. Furthermore, the valve control device 1350 transmits time series data of the control signals of the first valve 1320a and the second valve 1320b to the information processing device 230.

The first imaging device 220a captures images of the water to be treated Wa1 passing through the inside of the flow cell 1330. The first imaging device 220a extracts images at predetermined time intervals using an image extraction means (not shown) and transmits the images to the information processing device 230.

Based on the image received from the first imaging device 220a and the time series data of the control signal of the electromagnetic valve, the information processing device 230 distinguishes whether the image is an image of the treated water Wa1 sampled by the first water sampling pump (hereinafter referred to as the "first water sampling image") or an image of the treated water Wa1 sampled by the second water sampling pump (hereinafter referred to as the "second water sampling image").

The information storage means 231, like the first monitoring system, holds (memorizes, stores) treated water data, treated water data, first water sampling image data including a plurality of first water sampling images at a predetermined time interval, and second water sampling image data including a plurality of second water sampling images at a predetermined time interval in association with each other based on time information.

<Outline of operation>
The model generation means 232 of the second monitoring system, like the first monitoring system, uses the data stored in the information storage means 231 to generate a model for determining the success or failure of coagulation (a model that outputs information indicating the success or failure of the coagulation state of the turbidity in the treated water Wa1 shown in the image).

The model generating means 232 generates the first coagulation pass/fail judgment model and the second coagulation pass/fail judgment model in the same manner as the first monitoring system, except that the first water sample image data is used instead of the first image data and the second water sample image data is used instead of the second image data.

The coagulation quality determination means 233 uses either the first coagulation quality determination model or the second coagulation quality determination model generated by the model generation means 232 to calculate information indicating the quality of the coagulation state of the turbidity in the treated water Wa1 from input data including an image (the first water sample image or the second water sample image) showing the coagulation state of the turbidity in the treated water Wa1 at the current time. The coagulation quality determination means 233 is configured to be able to set whether to use the first coagulation quality determination model or the second coagulation quality determination model based on a user's operation input to an operation device not shown.

<Specific operation>
The model generating means 232 executes a process flow similar to that of Fig. 5, except that in the process flow of Fig. 5, the first image data is replaced with the first water sampled image data. The model generating means 232 executes a process flow similar to that of Fig. 6, except that in the process flow of Fig. 6, the second image data is replaced with the second water sampled image data. The coagulation quality determining means 233 executes a process flow similar to that of Fig. 7, except that in the process flow of Fig. 7, the first image is replaced with the first water sampled image. The coagulation quality determining means 233 executes a process flow similar to that of Fig. 8, except that in the process flow of Fig. 8, the second image is replaced with the second water sampled image. Note that these process flows are similar to the process flows of Figs. 5 to 8 except for the above points, and detailed explanations will be omitted.

＜Effects＞
As described above, the second monitoring system can capture images of the flocculation state of turbidity in the treated water Wa1 that exists in two locations (multiple locations) where the treated water Wa1 has been treated differently in a water purification plant, using a single first imaging device 220a. Furthermore, in the second monitoring system, the location where the flow cell 1330 is installed can be selected to a certain degree of freedom, so that the flow cell 1330 can be placed in a stable environment, such as indoors. Therefore, compared to the first monitoring system, the second monitoring system can improve the freedom of the location and imaging conditions of the treated water Wa1 that the first imaging device 220a wants to image.

<<Third embodiment>>
A water treatment status monitoring system according to a third embodiment of the present invention (hereinafter, may be referred to as a "third monitoring system") will be described. The third monitoring system differs from the first monitoring system only in the following points.

In the third monitoring system, the model generation means 232 extracts information about flocs (hereinafter referred to as "floc information") from the image data, and generates a model for determining whether the flocs are agglomerated or not, using the floc information as well. The floc information includes, for example, the number of flocs in the image, the size of the flocs (particle size, area, etc.), and the brightness of non-floc parts.

The following explanation will focus on these differences.

<Specific operation>
The model generating means 232 of the third monitoring system executes the process flow shown in Fig. 14 instead of the process flow shown in Fig. 5. Therefore, the model generating means 232 executes the processes of steps 1405 to 1430 described below in order, and then proceeds to step 1495 to end this process flow.

Step 1405: The model generation means 232 acquires data sets of the first image data, treated water data, and treated water data that are associated with each other from the information storage means 231.

Step 1410: The model generating means 232 performs image processing on each first image of the first image data. Specifically, the model generating means 232 grayscales each first image of the first image data and then performs brightness correction.

Step 1415: The model generation means 232 binarizes each first image of the first image data.

Step 1420: The model generation means 232 extracts flocks from each first image of the binarized first image data and obtains flock information for each first image.

Step 1425: The model generation means 232 normalizes the floc information and treated water data of each first image.

Step 1430: The model generation means 232 uses the processed first image data, the processed treated water data, the processed floc information, and the treated water data as learning data, and generates a learned model (first flocculation pass/fail judgment model) by deep learning, which takes the first image (first image after image processing), each measurement value of the treated water data (each measurement value after normalization processing), and the floc information (floc information after normalization processing) as input, and outputs predicted treated water turbidity. The model generation means 232 holds (stores) the generated first flocculation pass/fail judgment model.

The model generation means 232 of the third monitoring system executes the process flow shown in FIG. 15 instead of the process flow shown in FIG. 6. Therefore, the model generation means 232 executes the processes of steps 1505 to 1530 described below in order, and then proceeds to step 1595 to end this process flow.

Step 1505: The model generation means 232 acquires data sets of second image data, treated water data, and treated water data that are associated with each other from the information storage means 231.

Step 1510: The model generating means 232 performs image processing on each second image of the second image data. Specifically, the model generating means grayscales each second image of the second image data and then performs brightness correction.

Step 1515: The model generation means 232 binarizes each second image of the second image data.

Step 1520: The model generation means 232 extracts flocks from each second image of the binarized second image data and obtains flock information for each second image.

Step 1525: The model generation means 232 normalizes the floc information and treated water data of each second image.

Step 1530: The model generation means 232 uses the processed second image data, the processed treated water data, the processed floc information, and the treated water data as learning data, and generates, by deep learning, a trained model (second flocculation pass/fail judgment model) that takes as input the second image (second image after image processing), each measurement value of the treated water data (each measurement value after normalization processing), and the floc information (floc information after normalization processing) and outputs predicted treated water turbidity. The model generation means 232 retains (stores) the generated second flocculation pass/fail judgment model.

The flocculation quality determination means 233 executes a process flow that differs from the process flow in Fig. 7 only in the following respects. Note that, like the first monitoring system, this process flow is a process flow that is executed by the flocculation quality determination means 233 when the first flocculation quality determination model is selected as the model to be used for calculating information indicating the quality of the flocculation state of turbid matter.
Between step 710 and step 715, the following steps A1 and B1 are added.

Step A1: The aggregation quality determination means 233 binarizes the first image.

Step B1: The coagulation quality determination means 233 extracts flocs from the binarized first image and obtains floc information of the first image.
Instead of step 715, the following step C1 is executed, and instead of step 720, the following step D1 is executed.

Step C1: The flocculation quality determination means 233 normalizes the treated water data (each measurement value at the current time) and the floc information.

Step D1: The flocculation quality judgment means 233 acquires a first flocculation quality judgment model from the model generation means 232, inputs the first image that has been image-processed, the treated water data after normalization (each measured value after normalization), and the floc information (floc information after normalization) into the first flocculation quality judgment model, and calculates and acquires a predicted value of the treated water turbidity (predicted treated water turbidity), which is the output of the first flocculation quality judgment model.

The coagulation quality determining means 233 executes a process flow which differs from the process flow of FIG. 8 only in the following respects.
The following steps A2 and B2 are added between steps 810 and 815. Note that, like the second monitoring system, this processing flow is a processing flow executed by the coagulation quality determination means 233 when the second coagulation quality determination model is selected as the model to be used for calculating information indicating the quality of the coagulation state of turbid matter.

Step A2: The aggregation quality determination means 233 binarizes the second image.

Step B2: The coagulation quality determination means 233 extracts flocs from the binarized second image and obtains floc information of the second image.
Instead of step 815, the following step C2 is executed, and instead of step 820, the following step D2 is executed.

Step C2: The flocculation quality determination means 233 normalizes the treated water data (each measurement value at the current time) and the floc information.

Step D2: The flocculation quality judgment means 233 acquires the second flocculation quality judgment model from the model generation means 232, inputs the second image that has been image-processed, the treated water data after normalization (each measured value after normalization), and the floc information (floc information after normalization) into the second flocculation quality judgment model, and calculates and acquires a predicted value of the treated water turbidity (predicted treated water turbidity), which is the output of the second flocculation quality judgment model.

＜Effects＞
As described above, the third monitoring system generates a first coagulation quality judgment model and a second coagulation quality judgment model, also using floc information that affects the quality of the coagulation state of the turbidity in the treated water Wa1, and can more accurately calculate information indicating the quality of the coagulation state of the turbidity in the treated water Wa1 shown in the image using the generated coagulation quality judgment model (the first coagulation quality judgment model or the second coagulation quality judgment model).

<<Fourth embodiment>>
A water treatment status monitoring system according to a fourth embodiment of the present invention (hereinafter, may be referred to as a "fourth monitoring system") will be described. The fourth monitoring system differs from the first monitoring system only in the following points.

The fourth monitoring system generates a single aggregation pass/fail judgment model using the first image data and the second image data.

The following explanation will focus on these differences.

<Specific operation>
The model generating means 232 of the fourth monitoring system executes the processing flow shown in Fig. 16. Therefore, the model generating means 232 starts processing from step 600 in Fig. 16, executes the processing from step 1605 to step 1625 described below in order, and then proceeds to step 1695 to end this processing flow.

Step 1605: The model generation means 232 acquires data sets of the first image data, the second image data, the treated water data, and the treated water data that are associated with each other from the information storage means 231.

Step 1610: The model generating means 232 performs image processing on each first image of the first image data. Specifically, the model generating means 232 grayscales each first image of the first image data and then performs brightness correction.

Step 1615: The model generating means 232 performs image processing on each second image of the second image data. Specifically, the model generating means 232 grayscales each second image of the second image data and then performs brightness correction.

Step 1620: The model generation means 232 normalizes the treated water data (each measured value of the treated water data).

Step 1625: The model generation means 232 uses the processed first image data, the processed second image data, the treated water data, and the treated water data as learning data, and generates a learned model (coagulation pass/fail judgment model) by deep learning, using the first image (first image after image processing) and the second image (second image data after image processing) and each measured value of the treated water data (each measured value after normalization processing) as input, and outputting predicted treated water turbidity. The model generation means 232 holds (stores) the generated coagulation pass/fail judgment model.

Figure 17 is a flowchart showing the process flow executed by the coagulation quality determination means 233. The coagulation quality determination means 233 executes the flowchart shown in Figure 17. Therefore, the coagulation quality determination means 233 starts processing from step 1700 in Figure 17, and sequentially executes the processes of steps 1705 to 1720 described below. Thereafter, the coagulation quality determination means 233 proceeds to step 1795 and temporarily ends this process flow.

Step 1705: The flocculation quality determination means 233 acquires a first image at the current time from the first imaging device 220a, acquires a second image at the current time from the second imaging device 220b, and acquires treated water data (each measurement value at the current time) from the first water quality meter 210a.

Step 1710: The agglomeration quality determination means 233 processes the first image. Specifically, the agglomeration quality determination means 233 converts the first image into a grayscale image and then performs brightness correction.

Step 1715: The aggregation quality determination means 233 processes the second image. Specifically, the aggregation quality determination means 233 converts the second image into a grayscale image and then performs brightness correction.

Step 1720: The coagulation pass/fail determination means 233 normalizes the treated water data (each measured value at the current time).

Step 1725: The coagulation quality judgment means 233 acquires a coagulation quality judgment model from the model generation means 232, inputs the first image that has been image-processed and the treated water data that has been normalized into the first coagulation quality judgment model, and calculates and acquires a predicted value of the treated water turbidity (predicted treated water turbidity), which is the output of the first coagulation quality judgment model.

＜Effects＞
As described above, the fourth monitoring system generates one coagulation quality judgment model using the first image data and the second image data of the treated water Wa1 located at different locations, which reflect the formation of coagulation nuclei and flocs (changes over time), and calculates information indicating the quality of the coagulation state of the turbidity using the generated coagulation quality judgment model. This allows the fourth monitoring system to accurately calculate information indicating the quality of the coagulation state of the turbidity in the treated water Wa1 shown in the image. Furthermore, the fourth monitoring system can reduce parameters related to model selection compared to the first monitoring system, making it possible to simplify the system.

According to the above-described embodiments and the first modified example, for example, in developing countries where there are no skilled workers, it is possible to perform appropriate water treatment, thereby improving the hygiene of people living in those areas, and contributing to the realization of a sustainable society through environmental conservation through the use of appropriate chemicals.

<<Other variations>>
The present invention is not limited to the above-described embodiments and the first modification, and various modifications can be adopted within the scope of the present invention. The above-described embodiments and the first modification can be combined with each other within the scope of the present invention.

For example, the first embodiment and the fourth embodiment may be combined. In this case, for example, the model generating means 232 generates a plurality of coagulation quality judgment models (in this example, the first coagulation quality judgment model, the second coagulation quality judgment model, and the coagulation quality judgment model). The coagulation quality judgment means 233 uses any of the plurality of coagulation quality judgment models (in this example, the first coagulation quality judgment model, the second coagulation quality judgment model, and the coagulation quality judgment model) generated by the model generating means 232 to calculate information indicating the quality of the coagulation state of the turbidity of the water to be treated Wa1 from input data including an image showing the coagulation state of the turbidity of the water to be treated Wa1.

For example, in each of the above embodiments and the first modified example, the model generation means 232 generated a learned model by deep learning as a model for determining whether or not turbidity in the treated water Wa1 has flocculated, but the model for determining whether or not flocculation has occurred may generate an empirical formula, a statistical model, or the like, as long as it is capable of indicating whether or not flocculation has occurred.

For example, in each of the above embodiments and the first modified example, the imaging conditions for the image of the treated water Wa1 are not particularly limited as long as the flocculation state of the turbid matter in the treated water Wa1 can be imaged. However, if the only light source is sunlight and the brightness differs greatly between day and night, it is preferable to install a light source that illuminates the imaging location of the imaging device 220. Furthermore, since recording the change in brightness and using it for pre-processing (brightness correction) of the image of the treated water Wa1 will improve the accuracy of determining whether the flocculation state of the turbid matter is good or bad, a light meter or the like may be installed to record the change in brightness of the imaging location.

In each of the first to third embodiments and the first modified example, for example, if the water quality of the water to be treated Wa1 is relatively stable and does not fluctuate easily, it is preferable to investigate in advance the treatment time (or location) within the water purification plant where differences in the flocculation state of turbidity are likely to occur, and to install the imaging device 220 (in the second embodiment, a water sampling pump) in a location where the water to be treated Wa1 where differences in the flocculation state of turbidity are likely to occur can be imaged (in the second embodiment, a location where water can be sampled).

In each of the above first to third embodiments and the first modified example, for example, when the water quality of the water to be treated Wa1 varies, the image (input data) and the coagulation quality judgment model used by the information processing device 230 to judge the quality of the coagulation may be switched manually or automatically as appropriate depending on the water quality of the water to be treated Wa1 (based on the water to be treated data).

In this case, for example, the information processing device 230 may determine, based on the treated water data, whether the treated water data satisfies the conditions that promote floc growth, and based on the determination result, if the treated water data satisfies the conditions that promote floc growth, select a coagulation quality judgment model generated using image data captured by an imaging device installed at an upstream location, compared to when the treated water Wa1 does not satisfy the conditions that promote floc growth, and use the selected coagulation quality judgment model to output a predicted treated water turbidity from input data including an image of the treated water Wa1.

For example, if the turbidity of the treated water Wa1 is approximately 80 degrees, the treated water data meets the conditions that promote the growth of flocs, and an image captured by an imaging device 220 installed in a position capable of capturing an image of the treated water Wa1 corresponding to the upstream side (e.g., 0 minute slow stirring processing time) is used to generate a coagulation quality judgment model, and a predicted treated water turbidity can be output from the image using the generated coagulation quality judgment model.

For example, if the turbidity of the treated water Wa1 is approximately 35 degrees, the treated water data does not meet the conditions that promote the growth of flocs, and an image captured by an imaging device 220 installed in a position capable of capturing an image of the treated water Wa1 corresponding to the downstream side (e.g., a processing time of 4 minutes with slow stirring) can be used to generate a coagulation quality judgment model, and a predicted treated water turbidity can be output from the image using the generated coagulation quality judgment model.

In each of the above embodiments and the first modified example, three or more imaging devices 220 may be installed in different locations, and multiple coagulation quality judgment models may be generated using image data captured by each of the imaging devices 220. In this case, as in the case where there are two coagulation quality judgment models as in each of the above first to third embodiments and the first modified example, one coagulation quality judgment model may be selected from the multiple coagulation quality judgment models, and the selected coagulation quality judgment model may be used to output a predicted treated water turbidity from an image of the treated water Wa1.

In the above four embodiments, three or more imaging devices 220 may be installed at different locations, and the information processing device 230 may generate a single aggregation pass/fail judgment model using image data captured by each imaging device 220.

In each of the above embodiments and the first modified example, the information processing device 230 may perform a binary classification by outputting either information indicating the good or bad flocculation state of the turbid matter as output from the flocculation quality determination model.

In each of the above embodiments and the first modified example, the information processing device 230 may determine whether the flocculation state of the turbidity is good or bad by comparing the predicted treated water turbidity output from the flocculation good/bad judgment model with a predetermined treated water threshold turbidity (e.g., 1 degree) and output the judgment result.

In each of the above embodiments and the first modified example, the information processing device 230 may adjust the amount of coagulant injected from the chemical pump via the monitoring control device according to the predicted treated water turbidity. For example, when the coagulation state is poor (not poor), the information processing device 230 may adjust the amount of coagulant injected to make the coagulation state appropriate and adjust the turbidity of the treated water Wa2 so that it is clean.

In each of the above embodiments and the first modified example, the information processing device 230 may generate an aggregation pass/fail model using only image data.

The above embodiments and the first modified example are applied to the water purification plant 100, but may also be applied to other plants that purify water other than the water purification plant 100.

In the first embodiment described above, the information processing device 230 may select at least two models (in this example, the first coagulation quality judgment model and the second coagulation quality judgment model) to be used in calculating information indicating the quality of the coagulation state of the turbidity matter, and may calculate the final predicted treated water turbidity by multiplying the predicted treated water turbidity output from each of the at least two coagulation quality judgment models (in this example, the first coagulation quality judgment model and the second coagulation quality judgment model) by a weight previously set according to the turbidity of the water to be treated, and adding them together.

In each of the above embodiments and the first modified example, the information processing device 230 may output the result determined by the flocculation quality determination means 233 to a monitor or the like connected to the information processing device 230. In this case, for example, the information processing device 230 may be connected to a device that issues an alarm, and the information processing device 230 may cause the alarm device to issue an alarm if the predicted treated water turbidity exceeds a threshold value (1 degree).

Even if there is only one imaging device 220, if it can move along the water treatment flow (rail type, drone, etc.), one device can capture images of multiple locations. In this case, the present invention can be realized by recording information that can identify the imaging location together with the image data. One method of obtaining information that can identify the imaging location is to add a GPS (Global Positioning System) function to the imaging device 220 and acquire GPS information, for example.

In each of the above embodiments and the first modified example, the information processing device 230 may generate a coagulation pass/fail judgment model in which input data such as an image at a certain time and/or treated water data (each measured value) is replaced with an image within a certain time range and/or treated water data.

In each of the above embodiments and the first modified example, when the water to be treated is treated sequentially at one location without being transported to multiple treatment locations, it is possible to capture images of the progress of the treatment status (multiple locations with different treatment times) with one device. In this case, the model generation means 232 may generate a flocculation pass/fail judgment model for each of the multiple locations with different treatment times.

100... water purification plant, 210a... first water quality meter, 210b... second water quality meter, 220a... first imaging device, 220b... second imaging device, 230... information processing device, 231... information storage means, 232... model generation means, 233... flocculation quality determination means, 234... flocculation quality output means

Claims (14)

A water treatment status monitoring system applied to a facility that produces purified treated water by transporting the water to be treated to a plurality of treatment locations in sequence and performing a process including a coagulation and sedimentation process for removing turbidity contained in the water to be treated,
A plurality of imaging devices are installed at a plurality of different locations capable of imaging the water to be treated, and image the water to be treated to obtain image data of the water to be treated for each location, including images of the water to be treated captured from each location;
a water measurement device for measuring water quality parameters including turbidity of the water to be treated and acquiring water data including the water quality parameters of the water to be treated;
a treated water measuring device for measuring the turbidity of the treated water and acquiring treated water data including the turbidity of the treated water;
an information processing device that acquires the image data, the treated water data, and the treated water data from the imaging device, the treated water measuring device, and the treated water measuring device;
Equipped with
The information processing device includes:
The learning input data including the image data of the water to be treated and the treated water data are used as learning data, and a plurality of coagulation quality judgment models are created by machine learning, which output information indicating the quality of the coagulation state of the turbidity of the water to be treated from the input data including the image of the water to be treated,
The information processing device includes:
Based on the turbidity of the treated water,
Selecting at least one of the plurality of flocculation quality judgment models, acquiring the image of the water to be treated from at least one of the plurality of imaging devices as an image for judgment, and using the selected flocculation quality judgment model, calculating information indicating the quality of the flocculation state of the turbidity of the water to be treated shown in the image for judgment from judgment input data including the image for judgment.
It was configured as follows:
Water treatment status monitoring system. The water treatment status monitoring system according to claim 1,
the information processing device creates one of the agglutination quality determination models from the image data captured by one of the imaging devices;
It was configured as follows:
Water treatment status monitoring system. The water treatment status monitoring system according to claim 1,
the information processing device creates one of the agglutination quality determination models from the image data captured by the multiple imaging devices;
It was configured as follows:
Water treatment status monitoring system. A water treatment status monitoring system applied to a facility that produces purified treated water by transporting the water to be treated to a plurality of treatment locations in sequence and performing a process including a coagulation and sedimentation process for removing turbidity contained in the water to be treated,
A plurality of imaging devices are installed at a plurality of different locations capable of imaging the water to be treated, and image the water to be treated to obtain image data of the water to be treated for each location, including images of the water to be treated captured from each location;
a water measurement device for measuring water quality parameters including turbidity of the water to be treated and acquiring water data including the water quality parameters of the water to be treated;
a treated water measuring device for measuring the turbidity of the treated water and acquiring treated water data including the turbidity of the treated water;
an information processing device that acquires the image data, the treated water data, and the treated water data from the imaging device, the treated water measuring device, and the treated water measuring device;
An operation device connected to the information processing device,
The information processing device includes:
The learning input data including the image data of the water to be treated and the treated water data are used as learning data, and a plurality of coagulation quality judgment models are created by machine learning, which output information indicating the quality of the coagulation state of the turbidity of the water to be treated from the input data including the image of the water to be treated,
The information processing device selects at least one of the plurality of coagulation quality judgment models based on an operation on the operation device, acquires the image of the water to be treated from at least one of the plurality of imaging devices as an image for judgment, and calculates information indicating the quality of the coagulation state of the turbidity of the water to be treated shown in the image for judgment from input data for judgment including the image for judgment using the selected coagulation quality judgment model.
It was configured as follows:
Water treatment status monitoring system. The water treatment status monitoring system according to claim 1 ,
the information processing device compares the turbidity of the treated water with a threshold turbidity, and selects one of the multiple flocculation pass/fail judgment models based on the comparison result.
It was configured as follows:
Water treatment status monitoring system. The water treatment status monitoring system according to claim 5 ,
the information processing device has a storage unit that stores a data set including the image data, the to-be-treated water data, and the treated water data;
The information processing device uses a portion of the data set stored in the memory unit as the learning data to create a plurality of the flocculation quality judgment models, and calculates the threshold turbidity by evaluating the accuracy of information indicating the quality of the flocculation state of the turbidity of the treated water output by the created plurality of the flocculation quality judgment models using at least a portion of the remaining portion excluding the portion of the data set, according to the turbidity of the treated water.
It was configured as follows:
Water treatment status monitoring system. A water treatment status monitoring system applied to a facility that produces purified treated water by transporting the water to be treated to a plurality of treatment locations in sequence and performing a process including a coagulation and sedimentation process for removing turbidity contained in the water to be treated,
A plurality of imaging devices are installed at a plurality of different locations capable of imaging the water to be treated, and image the water to be treated to obtain image data of the water to be treated for each location, including images of the water to be treated captured from each location;
a water measurement device for measuring water quality parameters including turbidity of the water to be treated and acquiring water data including the water quality parameters of the water to be treated;
a treated water measuring device for measuring the turbidity of the treated water and acquiring treated water data including the turbidity of the treated water;
an information processing device that acquires the image data, the treated water data, and the treated water data from the imaging device, the treated water measuring device, and the treated water measuring device;
Equipped with
The information processing device includes:
The learning input data including the image data of the water to be treated and the treated water data are used as learning data, and a plurality of coagulation quality judgment models are created by machine learning, which output information indicating the quality of the coagulation state of the turbidity of the water to be treated from the input data including the image of the water to be treated,
The information processing device includes:
Selecting at least one of the plurality of flocculation quality judgment models, acquiring the image of the water to be treated from at least one of the plurality of imaging devices as an image for judgment, and using the selected flocculation quality judgment model, calculating information indicating the quality of the flocculation state of the turbidity of the water to be treated shown in the image for judgment from judgment input data including the image for judgment.
It is configured as follows:
The information processing device calculates a predicted turbidity of the treated water as information indicating whether the flocculation state of the turbidity of the treated water shown in the determination image is good or bad.
It was configured as follows:
Water treatment status monitoring system. A water treatment status monitoring system applied to a facility that produces purified treated water by transporting the water to be treated to a plurality of treatment locations in sequence and performing a process including a coagulation and sedimentation process for removing turbidity contained in the water to be treated,
A plurality of imaging devices are installed at a plurality of different locations capable of imaging the water to be treated, and image the water to be treated to obtain image data of the water to be treated for each location, including images of the water to be treated captured from each location;
a water measurement device for measuring water quality parameters including turbidity of the water to be treated and acquiring water data including the water quality parameters of the water to be treated;
a treated water measuring device for measuring the turbidity of the treated water and acquiring treated water data including the turbidity of the treated water;
an information processing device that acquires the image data, the treated water data, and the treated water data from the imaging device, the treated water measuring device, and the treated water measuring device;
Equipped with
The information processing device includes:
The learning input data including the image data of the water to be treated and the treated water data are used as learning data, and a plurality of coagulation quality judgment models are created by machine learning, which output information indicating the quality of the coagulation state of the turbidity of the water to be treated from the input data including the image of the water to be treated,
The information processing device includes:
Selecting at least one of the plurality of flocculation quality judgment models, acquiring the image of the water to be treated from at least one of the plurality of imaging devices as an image for judgment, and using the selected flocculation quality judgment model, calculating information indicating the quality of the flocculation state of the turbidity of the water to be treated shown in the image for judgment from judgment input data including the image for judgment.
It is configured as follows:
The information processing device calculates a judgment result of either good or bad as information indicating the good or bad flocculation state of the turbidity of the treated water reflected in the judgment image.
It was configured as follows:
Water treatment status monitoring system. A water treatment status monitoring system applied to a facility that produces purified treated water by transporting the water to be treated to a plurality of treatment locations in sequence and performing a process including a coagulation and sedimentation process for removing turbidity contained in the water to be treated,
A plurality of imaging devices are installed at a plurality of different locations capable of imaging the water to be treated, and image the water to be treated to obtain image data of the water to be treated for each location, including images of the water to be treated captured from each location;
a water measurement device for measuring water quality parameters including turbidity of the water to be treated and acquiring water data including the water quality parameters of the water to be treated;
a treated water measuring device for measuring the turbidity of the treated water and acquiring treated water data including the turbidity of the treated water;
an information processing device that acquires the image data, the treated water data, and the treated water data from the imaging device, the treated water measuring device, and the treated water measuring device;
Equipped with
The information processing device includes:
The learning input data including the image data of the water to be treated and the treated water data are used as learning data, and a plurality of coagulation quality judgment models are created by machine learning, which output information indicating the quality of the coagulation state of the turbidity of the water to be treated from the input data including the image of the water to be treated,
The information processing device includes:
Selecting at least one of the plurality of flocculation quality judgment models, acquiring the image of the water to be treated from at least one of the plurality of imaging devices as an image for judgment, and using the selected flocculation quality judgment model, calculating information indicating the quality of the flocculation state of the turbidity of the water to be treated shown in the image for judgment from judgment input data including the image for judgment.
It is configured as follows:
The learning input data includes only the image data of the water to be treated,
The input data for determination includes only the image for determination.
Water treatment status monitoring system. A water treatment status monitoring system applied to a facility that produces purified treated water by transporting the water to be treated to a plurality of treatment locations in sequence and performing a process including a coagulation and sedimentation process for removing turbidity contained in the water to be treated,
A plurality of imaging devices are installed at a plurality of different locations capable of imaging the water to be treated, and image the water to be treated to obtain image data of the water to be treated for each location, including images of the water to be treated captured from each location;
a water measurement device for measuring water quality parameters including turbidity of the water to be treated and acquiring water data including the water quality parameters of the water to be treated;
a treated water measuring device for measuring the turbidity of the treated water and acquiring treated water data including the turbidity of the treated water;
an information processing device that acquires the image data, the treated water data, and the treated water data from the imaging device, the treated water measuring device, and the treated water measuring device;
Equipped with
The information processing device includes:
The learning input data including the image data of the water to be treated and the treated water data are used as learning data, and a plurality of coagulation quality judgment models are created by machine learning, which output information indicating the quality of the coagulation state of the turbidity of the water to be treated from the input data including the image of the water to be treated,
The information processing device includes:
Selecting at least one of the plurality of flocculation quality judgment models, acquiring the image of the water to be treated from at least one of the plurality of imaging devices as an image for judgment, and using the selected flocculation quality judgment model, calculating information indicating the quality of the flocculation state of the turbidity of the water to be treated shown in the image for judgment from judgment input data including the image for judgment.
It is configured as follows:
The information processing device acquires the water quality parameters of the treated water from the treated water measuring device as judgment water quality parameters of the treated water;
The learning input data further includes the treated water data,
The input data for judgment further includes the water quality parameter for judgment.
Water treatment status monitoring system. A water treatment status monitoring system applied to a facility that produces purified treated water by transporting the water to be treated to a plurality of treatment locations in sequence and performing a process including a coagulation and sedimentation process for removing turbidity contained in the water to be treated,
A plurality of imaging devices are installed at a plurality of different locations capable of imaging the water to be treated, and image the water to be treated to obtain image data of the water to be treated for each location, including images of the water to be treated captured from each location;
a water measurement device for measuring water quality parameters including turbidity of the water to be treated and acquiring water data including the water quality parameters of the water to be treated;
a treated water measuring device for measuring the turbidity of the treated water and acquiring treated water data including the turbidity of the treated water;
an information processing device that acquires the image data, the treated water data, and the treated water data from the imaging device, the treated water measuring device, and the treated water measuring device;
Equipped with
The information processing device includes:
The learning input data including the image data of the water to be treated and the treated water data are used as learning data, and a plurality of coagulation quality judgment models are created by machine learning, which output information indicating the quality of the coagulation state of the turbidity of the water to be treated from the input data including the image of the water to be treated,
The information processing device includes:
Selecting at least one of the plurality of flocculation quality judgment models, acquiring the image of the water to be treated from at least one of the plurality of imaging devices as an image for judgment, and using the selected flocculation quality judgment model, calculating information indicating the quality of the flocculation state of the turbidity of the water to be treated shown in the image for judgment from judgment input data including the image for judgment.
It is configured as follows:
The information processing device is configured to acquire learning floc data including floc information that is information about flocs that are lumps of turbidity in each image of the image data of the water to be treated, and to acquire determination floc information that is information about flocs that are lumps of turbidity from the determination image,
The learning input data further includes the learning block data,
The input data for judgment further includes the flock information for judgment.
Water treatment status monitoring system. A water treatment status monitoring system applied to a facility that produces purified treated water by transporting the water to be treated to a plurality of treatment locations in sequence and performing a process including a coagulation and sedimentation process for removing turbidity contained in the water to be treated,
A plurality of imaging devices are installed at a plurality of different locations capable of imaging the water to be treated, and image the water to be treated to obtain image data of the water to be treated for each location, including images of the water to be treated captured from each location;
a water measurement device for measuring water quality parameters including turbidity of the water to be treated and acquiring water data including the water quality parameters of the water to be treated;
a treated water measuring device for measuring the turbidity of the treated water and acquiring treated water data including the turbidity of the treated water;
an information processing device that acquires the image data, the treated water data, and the treated water data from the imaging device, the treated water measuring device, and the treated water measuring device;
Equipped with
The information processing device includes:
The learning input data including the image data of the water to be treated and the treated water data are used as learning data, and a plurality of coagulation quality judgment models are created by machine learning, which output information indicating the quality of the coagulation state of the turbidity of the water to be treated from the input data including the image of the water to be treated,
The information processing device includes:
Selecting at least one of the plurality of flocculation quality judgment models, acquiring the image of the water to be treated from at least one of the plurality of imaging devices as an image for judgment, and using the selected flocculation quality judgment model, calculating information indicating the quality of the flocculation state of the turbidity of the water to be treated shown in the image for judgment from judgment input data including the image for judgment.
It is configured as follows:
The information processing device includes:
determining whether the data on the water to be treated corresponds to a condition that promotes the growth of flocs, which are masses of turbidity, based on the water quality parameters of the water to be treated;
When the water quality parameters of the treated water satisfy the conditions for promoting the growth of the flocs, the flocculation quality judgment model created using the image data captured at an upstream location is selected, compared to when the water quality parameters of the treated water do not satisfy the conditions for promoting the growth of the flocs.
It was configured as follows:
Water treatment status monitoring system. A water treatment status monitoring system applied to a facility that produces purified treated water by transporting the water to be treated to a plurality of treatment locations in sequence and performing a process including a coagulation and sedimentation process for removing turbidity contained in the water to be treated,
A plurality of water sampling devices that sample the water to be treated, which are located in a plurality of different locations;
A plurality of imaging devices that capture images of the water to be treated that existed in a plurality of locations and that were sampled by the plurality of water sampling devices, thereby acquiring sampled image data of the water to be treated for each location, including sampled images that are images of the water to be treated that existed in each of the locations;
a water measurement device for measuring water quality parameters including turbidity of the water to be treated and acquiring water data including the water quality parameters of the water to be treated;
a treated water measuring device for measuring the turbidity of the treated water and acquiring treated water data including the turbidity of the treated water;
an information processing device that acquires the water sampling image data, the water to be treated data, and the treated water data from the imaging device, the water to be treated measuring device, and the treated water measuring device;
Equipped with
The information processing device includes:
The learning input data including the sampled image data of the water to be treated and the treated water data are used as learning data, and a flocculation quality judgment model that outputs information indicating the quality of the flocculation state of the turbidity of the water to be treated from the input data including the sampled image of the water to be treated by machine learning is created for each of the sampled image data of the water to be treated at each location,
The information processing device includes:
At least one of the multiple coagulation quality judgment models created for each of the sampled image data of the water to be treated corresponding to each location is selected based on the turbidity of the water to be treated , the sampled image of the water to be treated is acquired from the imaging device as an image for judgment, and information indicating the quality of the coagulation state of the turbidity of the water to be treated shown in the image for judgment is calculated from judgment input data including the image for judgment using the selected coagulation quality judgment model.
It was configured as follows:
Water treatment status monitoring system. A water treatment status monitoring method applied to a facility that produces purified treated water by transporting the water to be treated to a plurality of treatment locations in sequence and performing a process including a coagulation and sedimentation process for removing turbidity contained in the water to be treated, comprising:
By capturing images of the water to be treated using a plurality of imaging devices installed at a plurality of different locations capable of capturing images of the water to be treated, image data of the water to be treated for each location including images of the water to be treated captured from each location is obtained;
measuring water quality parameters including turbidity of the treated water with a water measuring device, and acquiring water quality parameters of the treated water including the water quality parameters of the treated water;
measuring the turbidity of the treated water by a treated water measuring device and acquiring treated water data including the turbidity of the treated water;
acquiring the image data, the treated water data, and the treated water data from the imaging device, the treated water measuring device, and the treated water measuring device by an information processing device;
Including,
By the information processing device,
Using the learning input data including the image data of the water to be treated and the treated water data as learning data, a plurality of flocculation quality judgment models are created by machine learning, which output information indicating the quality of the flocculation state of the turbidity of the water to be treated from the input data including the image of the water to be treated,
At least one of the plurality of coagulation quality judgment models is selected based on the turbidity of the treated water , the image of the treated water is acquired from at least one of the plurality of imaging devices as an image for judgment, and information indicating the quality of the coagulation state of the turbidity of the treated water reflected in the image for judgment is calculated from judgment input data including the image for judgment using the selected coagulation quality judgment model.
How to monitor water treatment status.

Images