JP7585137B2 - Water purification plant operation planning device, water purification plant operation planning method, and water purification plant operation management system - Google Patents

Description

The present invention relates to a water purification plant operation planning device, a water purification plant operation planning method, and a water purification plant operation management system.

A water purification plant takes water from a water source, performs various purification processes, and supplies the water to consumers. In this case, the water is temporarily stored in a water reservoir depending on the water demand of the consumers. Pipes are installed between these facilities in the water purification plant, and the water in the water purification plant is supplied to consumers through these pipes. Furthermore, each facility must be managed and operated according to various compliance level constraints (e.g. water level constraints) that are determined by design or operation.

Because water purification plants have such complex structures, workers have accumulated operational know-how over many years to grasp water demand, the state of each piece of equipment, and the weather, and to decide what treatment should be carried out accordingly (pump operation and pipe flow settings). However, collecting such operational know-how from workers, i.e., knowledge of what treatment should be carried out depending on the state of each piece of equipment at the water purification plant, has become more difficult in recent years due to the decrease in workers and the aging of the workforce.

Given this situation, for example, Patent Document 1 discloses a technology that mathematically solves the estimations and equipment within a water purification plant to calculate an operation plan. One of the aims of this technology is to appropriately reflect the individual conditions and operating know-how of each plant.

JP 2018-185678 A

However, with the technology of Patent Document 1, the operational know-how that can be taken into account is limited to the operating range of the equipment. For example, it is not possible to incorporate know-how according to the situation, such as operating in a certain way under certain conditions. In addition, because the equipment constraints are set as fixed conditions, factors such as the level of compliance with the constraints cannot be taken into account. Furthermore, because the determination of whether the operational know-how has been incorporated into the operation plan is made based on a simple difference between past operating results and the proposed operation plan, there are cases in which the operational know-how is not fully reflected.

The present invention was made in consideration of these problems, and its purpose is to provide a water purification plant operation planning device, a water purification plant operation planning method, and a water purification plant operation management system that are capable of creating operation plans that accurately reflect the operational know-how of water purification plant workers.

One aspect of the present invention to solve the above problem is a water purification plant operation planning device that includes an operation condition extraction unit that has a processor and memory, acquires a combination of the state of water in each piece of equipment at a water purification plant and the operation of each piece of equipment performed by an operator, and calculates operation conditions that are the relationship between the state of water in each piece of equipment at the water purification plant and the operation of each piece of equipment performed by an operator in response to that state based on the acquired combination, and an operation plan creation unit that calculates the operation that an operator will perform on each piece of equipment at the water purification plant under the operation conditions at a specified time point in the future that best satisfies the constraints of each piece of equipment at the water purification plant.

One aspect of the present invention for solving the above problem is a method for planning and developing operations at a water purification plant, in which an information processing device acquires a combination of the state of water in each piece of equipment at the water purification plant and the operations of each piece of equipment performed by an operator, and, based on the acquired combination, executes an operation condition extraction process to calculate an operation condition that is a relationship between the state of water in each piece of equipment at the water purification plant and the operations of each piece of equipment performed by an operator in response to that state, and an operation plan creation process to calculate an operation that an operator will perform on each piece of equipment at the water purification plant under the operation conditions at a specified time point in the future that best satisfies the constraints of each piece of equipment at the water purification plant.

One aspect of the present invention for solving the above problem is a water purification plant operation management system that includes an operational condition extraction unit having a processor and memory, which acquires a combination of the state of water in each piece of equipment at a water purification plant and the operation of each piece of equipment performed by an operator, and calculates operational conditions that are the relationship between the state of water in each piece of equipment at the water purification plant and the operation of each piece of equipment performed by an operator in response to that state, based on the acquired combination; an operational planning unit that calculates an operation to be performed on each piece of equipment at the water purification plant that best satisfies the constraints of each piece of equipment at the water purification plant under the operational conditions at a specified time in the future, and transmits a signal instructing the calculated operation; and a monitoring and control device that receives the signal and controls each piece of equipment based on the received signal.

According to the present invention, an operation plan can be created that accurately reflects the operational know-how of workers at a water purification plant.
Problems, configurations and effects other than those described above will become apparent from the following description of the embodiments.

FIG. 1 is a diagram illustrating an example of the configuration of a water purification plant operation management system. FIG. 11 is a diagram illustrating an example of operational performance data. FIG. 4 is a diagram illustrating an example of facility data. FIG. 4 is a diagram illustrating an example of network data. FIG. 2 is a diagram illustrating an example of hardware included in a water purification plant operation planning device and a monitoring control device. FIG. 11 is a flow diagram illustrating an example of a water purification plant operation plan formulation process. FIG. 11 is a flow diagram illustrating details of an operation condition extraction process. FIG. 11 is a diagram illustrating an example of a method for searching for a similar date. FIG. 13 is a diagram illustrating an example of conversion to learning data. FIG. 13 is a diagram illustrating an example of a data structure of operation conditions. FIG. 13 is a diagram illustrating an example of a display of operation conditions. FIG. 11 is a flow diagram illustrating details of an operation condition search process. FIG. 13 is a diagram illustrating an example of a method for calculating a confidence interval and an evaluation time of a water reservoir. 11 is a diagram illustrating an example of converting a derived operation condition into a constraint equation. FIG. FIG. 11 is a flow diagram illustrating details of a planning process. FIG. 13 is a diagram illustrating an example of constraints of an optimization model. FIG. 2 is a diagram illustrating an example of decision variables and an objective function of an optimization model. FIG. 13 is a diagram illustrating an example of data regarding explanatory variables used to generate an optimization model. FIG. 13 is a diagram illustrating an example of a demand amount calculated as a value of an explanatory variable by executing an optimization model. FIG. 13 is a diagram illustrating an example of an operation plan display screen.

1 is a diagram showing an example of the configuration of a water purification plant operation management system 1 according to this embodiment. The water purification plant operation management system 1 includes a monitoring control device 20 that acquires data on each facility from a water purification plant 40, a water purification plant operation plan making device 10 that makes an operation plan for each facility of the water purification plant 40 based on the data acquired by the monitoring control device 20, and an input/output device 30 that allows a user to input data to the water purification plant operation plan making device 10 or displays information to a user.

The water purification plant 40, the monitoring control device 20, the water purification plant operation planning device 10, and the input/output device 30 are connected to each other via, for example, a LAN (Local Area Network), a WAN (Wide Area Network), etc.
, and are communicatively connected via a wired or wireless network 5 such as the Internet or a dedicated line.

The water purification plant 40 has multiple systems 45. Each system 45 has various equipment such as a pump 41, a water purification device 43, a water distribution reservoir 44, and pipelines 42. The pipelines 42 connect each equipment. The pump 41 obtains water from the water source, distributes the water between each equipment, or provides water to consumers. The water distribution reservoir 44 temporarily stores water. The pump 41 is used to take water from the water source and distribute it between the equipment and to consumers.

The input/output device 30 has the functional units of an operational condition input unit 31 and a display unit 32. The operational condition input unit 31 accepts input of operational constraints (hereinafter referred to as operational conditions) for each piece of equipment in the water purification plant 40 from a user. The display unit 32 displays information output by the water purification plant operation planning device 10 on a screen. The input/output device 30 has input devices such as a keyboard, mouse, card reader, or touch panel, and output devices such as an LCD (Liquid Crystal Display), audio output device (speaker), or printer.

The monitoring and control device 20 acquires data on the status of each facility (e.g., start or stop status of the pump 41 and the water purification device 43, or data on the flow rate such as discharge rate and purified water amount, the flow rate and water level of the pipeline 42, and the water storage volume and water level of the water distribution reservoir 44) from each facility of the water purification plant 40 (pump 41, pipeline 42, water purification device 43, water distribution reservoir 44, etc.), and transmits the acquired data to the water purification plant operation planning device 10. The monitoring and control device 20 also controls each facility of the water purification plant 40 based on the operation plan created by the water purification plant operation planning device 10. For example, the monitoring and control device 20 controls the start or stop of the pump 41 and the water purification device 43, the flow rate and water level of the pipeline 42, and the water storage volume and water level of the water distribution reservoir 44.

Next, the water purification plant operation planning device 10 includes the functional units of a data collection unit 11, a demand forecasting unit 12, a constraint configuration unit 13, and an operation planning unit 14. The water purification plant operation planning device 10 also stores operation performance data 15, equipment data 16, network data 17, and environmental data 18, for example, in the form of a database.

The data collection unit 11 receives data sent from the monitoring control device 20 and stores the received data in the operational performance data 15.

The demand forecasting unit 12 predicts the amount of water demanded by consumers managed by the water purification plant 40.

The constraint configuration unit 13 has a function of automatically generating operational conditions in addition to generating operational conditions based on user input from the input/output device 30. That is, the constraint configuration unit 13 includes an operational condition extraction unit 131 that generates operational conditions.

The operating condition extraction unit 131 identifies the operating conditions, which are the relationship between the water state in each facility of the water purification plant 40 (e.g., the water level in the water reservoir 44) and the operation of each facility performed by workers in response to that state (e.g., starting or stopping the pump 41, adjusting the flow rate in the pipeline 42).

Based on these operating conditions, the display unit 32 of the input/output device 30 displays information on a plan display screen (described later) that shows the range of water conditions in each piece of equipment at the water purification plant 40 as "conditions" and the operations of each piece of equipment performed by workers in response to those conditions as "results."

The operation plan creation unit 14 calculates the operations that workers should perform on each piece of equipment in the water purification plant 40 that best satisfy the constraints of each piece of equipment (such as the water level constraints of the water reservoir 44) in the planned date, which is the target date for creating the operation plan for the water purification plant 40, under the above operating conditions.

The operational condition extraction unit 131 has the following functional units: an operational condition learning unit 132, an operational condition search unit 133, a confidence interval determination unit 134, and an evaluation time calculation unit 135.

The operation condition learning unit 132 identifies the operation conditions using machine learning. That is, the operation condition learning unit 132 acquires past performance information (operation performance data 15) including the state of water in each piece of equipment at the water purification plant 40 and the operation of each piece of equipment performed by workers in response to that state, and identifies the operation conditions by machine learning the relationship between the state of water selected from the performance information and the operation of the equipment.

To explain in more detail, the operational performance data 15 has information on the water condition and water demand in each facility of the water purification plant 40. The operational condition learning unit 132 calculates the water condition and water demand in each facility of the water purification plant 40 on the planned date based on the demand prediction unit 12, and extracts from the operational performance data 15 information on the water condition and equipment operation at a past point in time when the water condition and demand are determined to be highly similar to the calculated water condition and demand.

Next, the operating condition search unit 133 reconstructs the water state of each facility of the water purification plant 40 under the operating conditions from the perspective of the water flow path (system 45), and generates derived operating conditions, which are new operating conditions based on the reconstructed information. The operating condition search unit 133 calculates the operations that workers should perform on each facility in the water purification plant 40 that best satisfy the constraints of each facility under each derived operating condition by constructing an optimization model for each facility. The operating condition search unit 133 then calculates the degree of deviation between the facility state calculated together with the operation of each specified facility by the optimization model and the operation of each facility that was actually performed in the past, and applies the derived operating condition with the smallest calculated degree of deviation to the operation plan creation unit 14.

The confidence interval determination unit 134 and the evaluation time calculation unit 135 identify time periods in which the confidence interval of the measured value of the water condition is less than a predetermined width based on information on the range of the water condition of each facility of the water purification plant 40 at each past timing (for example, information on the upper and lower limits of the water level that the distributing reservoir 44 can tolerate). The operating condition search unit 133 generates derived operating conditions based on this confidence interval.
Next, the operation performance data 15, the facility data 16, and the network data 17 will be described in detail.

(Operational performance data)
2 is a diagram showing an example of the operation performance data 15. The operation performance data 15 is a database having data items such as water levels 151 of each reservoir, operation times 152 of each pump, discharge flow rates 153 of each pump, flow rates 154 of each pipeline, water demand 155 in each system, weather 156 at the water purification plant 40, and time 157 indicated by these pieces of information. Note that the types of equipment and types of parameters described here are merely examples, and other types of equipment and their parameters may be included.

(Facility data)
3 is a diagram showing an example of the equipment data 16. The equipment data 16 is a database that stores information on absolute constraints (hard constraints) that each piece of equipment has, and relative constraints (soft constraints) that each piece of equipment has. In this embodiment, the hard constraints have design upper and lower limits as absolute constraints (e.g., physical constraints). The soft constraints have design upper and lower limits as absolute constraints, and operational upper and lower limits as constraints recommended for operation.

Specifically, the equipment data 16 includes the design upper limit 161 and the design lower limit 162, as well as the operational lower limit 163 and the operational lower limit 164, of the water level 165 of each reservoir, the design upper limit 161 and the design lower limit 162, as well as the operational lower limit 163 and the operational lower limit 164, of the discharge flow rate 166 of each pump, and the design upper limit 161 and the design lower limit 162, as well as the operational lower limit 163 and the operational lower limit 164, of the flow rate 167 of each pipeline. Note that the types of equipment and the types of parameters described here are merely examples, and other types of equipment and their parameters may be included.

(Network Data)
4 is a diagram showing an example of the network data 17. The network data 17 is a database that stores the configuration of each facility of the water purification plant 40. In this embodiment, the network data 17 represents each facility, such as a water distribution reservoir 44 or a water purification facility 43, as a vertex, a flow path (pipe 42) between the facilities as a side, a facility on the upstream side of the water in the pipe 42 as a starting point, and a facility on the downstream side of the water in the pipe 42 as an end point.

Specifically, the network data 17 includes the type 171 (vertex), start point 172, and end point 173 of each reservoir 174, the type 171, start point 172, and end point 173 of each pump 176, the type 171 (side), start point 172, and end point 173 of each pipeline 177, and the type 171 (vertex) of the demand amount 178 of each system.

5 is a diagram illustrating an example of hardware included in the water purification plant operation planning device 10 and the monitoring control device 20. The water purification plant operation planning device 10 and the monitoring control device 20 each include a processor 91 such as a CPU (Central Processing Unit), an MPU (Micro Processing Unit), or a GPU (Graphics Processing Unit), a main memory device 92 such as a ROM (Read Only Memory) or a RAM (Random Access Memory), and a hard disk drive (Hard Drive).
The computer 100 is equipped with an auxiliary storage device 93 such as a USB (Universal Serial Interface) module, a USB (Disk Drive), a flash memory, or an SSD (Solid State Drive), and a communication device 94 such as a network interface card (NIC), a wireless communication module, a USB (Universal Serial Interface) module, or a serial communication module.

The functions of the water purification plant operation planning device 10 and the monitoring control device 20 are realized by the processor 91 reading and executing a program stored in the main storage device 92 or the auxiliary storage device 93. The above-mentioned program can be recorded on a recording medium and distributed, for example.
Next, the process performed by the water purification plant operation management system 1 will be described.

＜Water purification plant operation planning＞

Figure 6 is a flow diagram illustrating an example of a water purification plant operation planning process that constrains the state of each piece of equipment in the water purification plant 40 and the corresponding operations of workers as derived operation conditions, and automatically creates a future operation plan for the water purification plant 40 based on the conditions. The water purification plant operation planning process is started, for example, when a user makes a specified input to the water purification plant operation planning device 10.

First, the data collection unit 11 of the water purification plant operation planning device 10 reads the operation performance data 15, the equipment data 16, the network data 17, and the environmental data 18 (s1). For example, the data collection unit 11 reads the record contents of the data of the operation performance data 15 at a specific time in the past (e.g., the past three years), and the record contents of the data of the environmental data 18 at a specific time in the past (e.g., the past three years).

Based on the data read in step s1, the operation condition extraction unit 141 executes operation condition extraction process s2, which identifies past days (similar days) on which the state of the water purification plant 40 is similar to the planned date, and generates operation conditions using the similar days. Details of operation condition extraction process s2 will be described later.

The operational condition search unit 142 executes an operational condition search process s3 to generate derived operational conditions, which are new operational conditions that can calculate an optimal operational plan, based on the operational conditions generated in the operational condition extraction process s2. Details of the operational condition search process s3 will be described later.

The operation condition search unit 142 determines whether a highly accurate operation plan was created for the operation plan creation result for any of the derived operation conditions generated in the operation condition search process s3 (i.e., whether the deviation between the operation plan creation result and past actual values is less than or equal to a predetermined value) (s4). If a highly accurate operation plan was created (s4: Y), the process of s6 described below is executed, and if a highly accurate operation plan was not created (s4: N), the operation condition search unit 142 repeats the operation condition extraction process s2 after relaxing the judgment threshold (described in detail later) for identifying similar days by a predetermined value.

On the other hand, when the processing of s1 is completed, the demand forecasting unit 12 calculates the water demand for each system 45 of the water purification plant 40 on the planned date (s5).

For example, the demand forecasting unit 12 calculates the water demand of each system 45 of the water purification plant 40 on the planned date by estimating the relationship between the water demand and weather based on the past water demand and past weather information in the operation performance data 15. In addition, the demand forecasting unit 12 calculates the water state of each facility of each system 45 on the planned date by estimating the relationship between the water demand and the water state of each facility based on the past water state of each facility (such as the water level of the water reservoir 44) and the water demand in the operation performance data 15. Note that such prediction methods are disclosed, for example, in JP 2020-201609 A, JP 2019-203287 A, and JP 2018-185678 A.

The operation plan creation unit 14 executes a planning process s6 to generate information (operation plan) on the operations that workers will perform on each piece of equipment and the status of each piece of equipment on the planned date, based on the water demand calculated in s5, the hard and soft constraints related to the operation of the water purification plant 40, the derived operation conditions identified in the operation condition search process s3, and the latest data of the operation performance data 15 read in s1. Details of the planning process s6 will be described later.

The display unit 32 displays the operation plan generated in the planning process s6 on the screen (s9). Based on the displayed operation plan, the worker checks the work content for the planned date (starting or stopping the pump 41, changing the flow rate of each pipeline 42, etc.).

The operation plan creation unit 14 may transmit signals indicating the contents of the operation plan generated in the plan creation process s6 (e.g., an instruction signal to start or stop the pump 41, or an instruction signal to control the flow rate of the pipeline 42) to the monitoring control device 20 (s8). The monitoring control device 20 controls each facility of the water purification plant 40 based on the received signals. For example, the monitoring control device 20 controls the start or stop of the pump 41, or the flow rate of each pipeline 42.
This is the end of the water purification plant operation plan formulation process. Each step of the process will be described in detail below.

<Operation condition extraction process s2>
FIG. 7 is a flow diagram illustrating the details of the operation condition extraction process s2.
The operational condition extraction unit 141 calls the demand prediction unit 12 to calculate the water demand for each system 45 on the planned date and the status of each facility on the planned date (the water level of each distribution reservoir 44, the discharge flow rate of each pump 41, the flow rate of each pipeline 42, etc.) (s201).

Based on the water demand and the state of each piece of equipment calculated in s201, the operational condition extraction unit 141 searches for past days (similar days) on which the water demand and the state of each piece of equipment are similar to the planned date, and obtains information on the water demand and the state of each piece of equipment on the similar days (similar day information) (s202).

For example, the operational condition extraction unit 141 sets midnight on the planned date as the reference time and acquires weather information at the reference time. The operational condition extraction unit 141 also acquires the water demand and the state of each piece of equipment at the reference time from the water demand and the state of each piece of equipment calculated in s201. The operational condition extraction unit 141 acquires data for each past day from the records of the operational performance data 15, and calculates the sum of squares of the differences between each piece of acquired data and the previously acquired data on the water demand, the state of each piece of equipment, and the weather. The operational condition extraction unit 141 determines as a similar day a past day for which the difference between these two calculated sums of squares is equal to or less than a predetermined value (judgment threshold). The sum of squares (sum of squared errors) here is expressed by the following formula, for example:

(Sum of squared errors)
= (Demand forecast value at 0:00 on the planning date - Demand amount at 0:00 on the specified past date) ² + (Demand forecast value at 1:00 on the planning date - Demand amount at 1:00 on the specified past date) ² +...
+ (initial water level of reservoir A on the planning date - initial water level of reservoir A on the specified past date) ² + (initial water level of reservoir B on the planning date - initial water level of reservoir B on the specified past date) ² +...
+ (Flow rate of pipeline A on the planned date - Flow rate of pipeline A on the specified past date) ² +...

Note that each term in this equation may be standardized by a predetermined value (for example, the maximum value for each piece of equipment) or may be multiplied by a predetermined coefficient (a coefficient between 0 and 1) to make the digits uniform.

Here, FIG. 8 is a diagram explaining an example of a method for searching for similar days. In the case of the figure, the difference between the predicted value of water demand on the planned date and the demand on "past day 1" and "past day 2" is calculated (reference number 2021). In addition, the difference between the predicted water level of each reservoir 44 and the predicted flow rate of each pipeline 42 on the planned date and the predicted water level of each reservoir 44 and the flow rate of each pipeline 42 on "past day 1" and "past day 2" is calculated (reference number 2022). In the case of the figure, "past day 1" is closer to a similar day than "past day 2" for all of the demand, water level of each reservoir 44, and flow rate of each pipeline 42. As a result, only "past day 1" is determined to be a similar day.

7, the operational condition extraction unit 141 determines whether the number of similar days found in s202 is less than or equal to a predetermined number (e.g., 30) (s203). If the number of similar days is less than or equal to the predetermined number (s203: Y), the operational condition extraction unit 141 executes processing in s204, and if the number of similar days exceeds the predetermined number (s203: N), the operational condition extraction unit 141 executes processing in s207.

In s204, the operation condition extraction unit 141 fixes some of the variables representing the state of each piece of equipment and sets them as constants (hard constraint values), and sets the rest as decision variables. Then, based on these settings, the operation condition extraction unit 141 calls the operation plan creation process s6, which will be described later, by setting the similar date as the target date for calculation (s205).
The combination of the value of the decision variable obtained by step S202 and the value of the variable set to a fixed value in step S204 is stored as new similar day information related to the similar day. After that, the process in step S207 is performed.

In s207, the operation condition extraction unit 141 converts each similar day information acquired in s202 or stored in s205 into learning data for the optimization model. In this embodiment, the operation condition extraction unit 141 converts the information into learning data in which the operation of the equipment (the operation of each pump 41 and the flow rate of each pipeline 42) is the objective variable and the water condition of each equipment (such as the water level of each water reservoir 44) is the explanatory variable. The data items of this learning data are the same as those of the operation performance data 15.

9 is a diagram for explaining an example of conversion to learning data. The learning data includes start learning data 1008 in which the start of each pump is used as an objective variable, and stop learning data 1009 in which the stop of each pump 41 is used as an objective variable. The start or stop of the pump 41 is, for example, started (1) if the discharge flow rate of the pump 41 at a certain time is a positive value and the discharge flow rate one unit time ago is 0, and stopped (0) if the discharge flow rate of the pump 41 at a certain time is 0 and the discharge flow rate one unit time ago is a positive value.
Let us assume that.

The start-up learning data 1008 uses the water state of each facility at each time 1001 on a similar day when the pump 41 is running as an explanatory variable. The explanatory variable x of the start-up learning data 1008 is represented by matrix data 1004 with each explanatory variable 1002 as a row and each time 1001 on a similar day as a column. The objective variable y of the start-up learning data 1008 is represented by vector data 1005 with the objective variable 1003, which is 1 when the pump 41 is running and 0 when the pump 41 is stopped, as a column. The explanatory variables 1002 are the water level of each water reservoir 44, the operating state of each pump 41, the rate and amount of change in water demand, the deviation of water demand (deviation from the average value), and the deviation of water demand from the value at the same time on the previous day.

The stop learning data 1009 uses the water state of each facility at each time 1001 on a similar day when the pump 41 is stopped as an explanatory variable. The explanatory variable x of the stop learning data 1009 is represented by matrix data 1006 with each explanatory variable 1002 as a row and each time 1001 on a similar day as a column. The objective variable y of the stop learning data 1009 is represented by vector data 1007 with the objective variable 1003, which is 1 when the pump 41 is stopped and 0 when the pump 41 is running, as a column.

Next, as shown in FIG. 7, the operation condition extraction unit 141 performs machine learning on the learning data generated in s207 to calculate the relationship between the explanatory variables and the target variable, and calculates the importance of each explanatory variable (s208). In this embodiment, the operation condition extraction unit 141 calculates the relationship and the importance by performing learning using a decision tree.

The operational condition extraction unit 141 determines whether the number of iterations of the machine learning performed in s208 has reached a predetermined number (s209). If the number of iterations has not reached the predetermined number (s209: N), the operational condition extraction unit 141 executes the process of s210, and if the number of iterations has reached the predetermined number (s209: Y), the operational condition extraction unit 141 executes the process of s211.

In s210, the operation condition extraction unit 141 removes explanatory variables with low importance values calculated in s208 (for example, explanatory variables with importance lower than a predetermined threshold, or a predetermined number of explanatory variables with low importance) from the learning data, and repeats the process of s207.

In s211, the operational condition extraction unit 141 stores the contents of the learning results calculated as a result of the processing up to s210 as operational conditions. The operational condition extraction unit 141 also displays the generated operational information on the screen. This ends the operational condition extraction processing s2.

Figure 10 is a diagram showing an example of the data structure of the operation conditions. The output 1100 is composed of nodes 1101 representing each piece of equipment, and each node 1101 includes information on the conditions 1102 and their thresholds 1103 related to the equipment, and information on the operation 1104 of the equipment. Each node 1101 has multiple branches 1105 (True, False) specified by the conditions 1102 and the thresholds 1103. The operation conditions 1106 are information in the "If...then" format that represents the operation of the equipment performed by the worker when the state of each piece of equipment and the water demand satisfy each condition related to the thresholds. The condition may be a condition related to a specific value, or a condition related to a time-series change or a deviation from a past value. In this way, the output 1100 can be information on a decision tree in which the state of the equipment is the condition and the operation of the equipment is the result.

Figure 11 is a diagram showing an example of the display of operational conditions. As shown in the figure, the operational conditions may be displayed in a tree format (reference number 1151) as a data structure of the operational conditions, or may be displayed in an "if-then" format (reference number 1152). They may also be displayed together with the operation plan described below. For example, they may be displayed in a natural language format together with the operation plan (reference number 1153), or a table showing the contents of the operation plan may be displayed with the parts related to the operational conditions highlighted (reference number 1154).

<Operation condition search process s3>
12 is a flow diagram illustrating the operation condition search process s3 in detail. The operation condition search unit 142 reads the similar day information acquired in the operation condition extraction process s2 (s301).

Then, the operational condition search unit 142 calculates the confidence interval for each distributing reservoir 44 (s302-s304). That is, first, the operational condition search unit 142 acquires the fluctuations in water level for each distributing reservoir 44 over a specified period of time in the past from the operational performance data 15 (s302). Then, based on the distribution of water levels for each distributing reservoir 44 in each time zone during this specified period, the operational condition search unit 142 identifies the time at which the difference between the upper and lower limits of the confidence interval for the water level during the specified period is equal to or less than a specified value (s303). The operational condition search unit 142 stores the time identified in s302 as the evaluation time (s304).

13 is a diagram for explaining an example of a method for calculating the confidence interval and evaluation time of the reservoir 44. First, the operation condition search unit 142 acquires data on the fluctuation of the water level of each reservoir 44 for each hour (0:00, 1:00, ...) on each past day from the operation performance data 15 (reference numeral 1301), and converts it into data on the water level of each reservoir 44 in each time period of each day (reference numeral 1302). Based on this converted data, the operation condition search unit 142 counts the upper and lower limit values of the water level of each reservoir 44 for each time period (reference numeral 1303). Based on this counted data, the operation condition search unit 142 specifies a time period 1305 in which the difference between the upper and lower limit values of the water level is equal to or less than a predetermined value as the evaluation time.
In this embodiment, the confidence interval is set based on the water level of the distributing reservoir 44, but the confidence interval may be set based on the water condition of other facilities.

Next, as shown in FIG. 12, the operational condition search unit 142 reads the operational conditions generated in the operational condition extraction process s2 (s305). The operational condition search unit 142 then generates one or more new operational conditions (derived operational conditions) by reconstructing the state of each piece of equipment in these operational conditions (contents following "if"; hereafter referred to as the condition part) and the operations performed by the worker (contents following "then"; hereafter referred to as the result part) (s306). The operational condition search unit 142 then converts each derived operational condition into a constraint equation (s306).

Specifically, first, the operation condition search unit 142 identifies the water flow path (specifically, the system 45) indicated by the result part. The operation condition search unit 142 extracts all the equipment (equipment conditions) belonging to the identified system 45 from the condition part. The operation condition search unit 142 generates all combinations including at least any of the extracted equipment, and sets each generated combination as a new condition part, and sets each combination of each of these condition parts and the result part as a derived operation condition.

For example, if there is an operating condition "if A AND B AND C then W" (A and C are conditions of equipment belonging to the system of water purification plant P, B is a condition of equipment belonging to the system of water purification plant Q, and W is the operation of equipment of water purification plant P), "if A AND C then W" is extracted, and based on this, "if A AND C then W", "if C then W", and "if C then W" are generated as new derived operating conditions, respectively. Note that when generating derived operating conditions, derived operating conditions may be generated in which the threshold values related to the operating conditions are changed.

Furthermore, FIG. 14 is a diagram illustrating an example of converting derived operational conditions into a constraint equation. The operational condition search unit 142 converts the contents of "if...then..." into a linear expression. For example, if each condition in the if clause is a condition on which of two values a certain variable can take (e.g., 0 or 1), the operational condition search unit 142 generates an inequality 3061 that indicates the range of possible combinations of the two values in each condition.

Also, for example, if each condition in the if clause is a condition on whether a certain variable is in a range above a threshold (for example, whether a water reservoir is above a specified water level), the operation condition search unit 142 generates an inequality 3062 that includes a constraint equation that is 1 if a certain variable exceeds the threshold and 0 if it does not exceed the threshold.

Similarly, for example, when each condition in the if clause is a condition on whether a certain variable falls within a range below a threshold (for example, whether a water reservoir is below a specified water level), the operation condition search unit 142 generates an inequality 3063 including a constraint equation that is 1 if a certain variable exceeds a threshold and 0 if it does not exceed the threshold.

Next, as shown in FIG. 12, the operation condition search unit 133 constructs, for each derived operation condition, an optimization model for calculating the operation of each piece of equipment (the operation of the pump 41 and the flow rate of the pipeline 42) that satisfies the constraints indicated by the hard constraints of the equipment data 16 and minimizes deviation from the soft constraints under the derived operation conditions generated in s306 (s307). In this embodiment, the operation condition search unit 133 constructs the optimization model using mixed integer programming. Details of the optimization model will be described later.

The operational condition search unit 142 selects one of the similar days identified in the operational condition extraction process s2, and calculates the values of the decision variables on that similar day by inputting the values of each variable (such as changes in demand) at the reference time (e.g., 0:00) of the selected similar day into the optimization model generated in s307 (s308).

The operational condition search unit 142 repeats the process of step s308 for all similar days (s309).

For each derived operating condition, the operating condition search unit 142 compares the value (planned value) of the decision variable calculated by the optimization model in s308 with the past actual value obtained from the operating performance data 15 (s310). That is, the operating condition search unit 142 compares the water level (planned value) of each distributing reservoir 44 calculated when calculating the value of the decision variable in s308 with the water level (actual value) of each distributing reservoir 44 at the evaluation time obtained from the operating performance data 15. For example, the operating condition search unit 142 calculates the squared error between the water level of each distributing reservoir 44 and the water level of each distributing reservoir 44 at the evaluation time.

Based on the result of s310, the operational condition search unit 142 identifies the derived operational condition and the corresponding constraint equation that have the smallest deviation between the planned value and the actual value (s311). This ends the operational condition search process s3.

<Planning process s6>
15 is a flow diagram illustrating details of the planning process s6. The operation planning unit 14 sets a reference time for the planning date (e.g., 0:00 on August 2, 2018) (s601), and acquires the operation performance data 15 for the most recent period before the reference time (s602).

Next, the operation planning unit 14 constructs an optimization model for calculating the operation details of each facility (flow rate of each pipeline 42 and start or stop of each pump 41, etc.) that minimizes the deviation from the soft constraints of each facility (upper and lower operational limits of the water level of each reservoir, and upper and lower operational limits of the flow rate of each pipeline) for each time after the reference time (t = 0:00, 1:00, 2:00, ...) under the hard constraints of each facility (upper and lower design limits), water balance, and water demand constraints, etc. (s603). Note that when using data from the previous time in the calculation related to t = 0:00, the operation planning unit 14 uses the data acquired in s602.

The operation plan formulation unit 14 constructs an optimization model by, for example, using a library package that executes a solution method for mixed integer programming. Note that a method for constructing such an optimization model is disclosed in, for example, JP 2018-185678 A.
Here, a specific example of the configuration of the optimization model will be described.

(Constraints of the optimization model)
16 is a diagram for explaining an example of constraints of the optimization model. The constraints include a constraint 1401 on the water flow rate of the pipeline by each pump when the pump is started or stopped at time t, a balance 1402 of the water flow rate of each facility (excluding the reservoir 44) at time t, a balance change 1403 of the water storage volume of each reservoir 44 from time t-1 to time t, a balance 1404 (demand satisfaction condition) between the flow rate from each pipeline 42 at time t and the corresponding water demand in the system 45, a hard constraint 1405 (design constraint) on the water level of each reservoir 44 at time t, a hard constraint 1406 (design constraint) on the water flow rate in each pipeline 42 at time t, a soft constraint 1407 (operational constraint) on the water level of each reservoir 44 at time t, and a soft constraint 1408 (design constraint) on the water flow rate in each pipeline 42 at time t.

(Decision variables and objective functions of the optimization model)
17 is a diagram illustrating an example of decision variables and an objective function of the optimization model. First, the decision variables are the flow rate 1409 of each pipeline at time t, the operation 1410 (start or stop) of each pump 41 at time t, and the water storage volume 1411 of each water reservoir 44 at time t.

The objective function 1412 of the optimization model is a function that minimizes the sum of the deviation of the water level of each reservoir 44 from the soft constraints (upper and lower operational limits) and the deviation of the flow rate of each pipeline from the soft constraints (upper and lower operational limits). Note that the soft constraints may be set to different values for each time t. In addition, the objective function may include other operational information, such as the amount of power consumed when the pump is started.

Next, FIG. 18 is a diagram showing an example of data related to explanatory variables used to generate an optimization model. As shown in the figure, the operation plan creation unit 14 creates a
Information on the water level of each distributing reservoir 44, the operating time of the pump of each distributing reservoir 44, the discharge flow rate of each pump 41, the flow rate of each pipeline 42, and the water demand in each system 45 at "1:00" is obtained.

Figure 19 is a diagram showing an example of a predicted value of the demand calculated as part of the operation plan by executing the optimization model. As shown in the figure, this demand is the demand of each system 45 at each time after the reference time ("August 1, 2018, 0:00-", "August 1, 2018, 1:00-", ..., "August 1, 2018, 23:00-").

Next, as shown in FIG. 15, the operation plan creation unit 14 calculates the state of each piece of equipment and the operation of each piece of equipment (i.e., the operation plan) by inputting the hardware constraints and the water demand on the planned date, etc., into the optimization model generated in s603 (s604). For example, the operation plan creation unit 14 calculates the discharge flow rate of each pump 41, the flow rate of each pipeline 42, and the water level of each reservoir 44 as explanatory variables, and calculates the operation of each pump 41 and the flow rate of each pipeline as objective variables.

Then, the operation plan creation unit 14 displays the result calculated in s604 on the operation plan display screen, or outputs the result (s605).

<Operation plan display screen>
20 is a diagram showing an example of an operation plan display screen 1500. The operation plan display screen 1500 includes an operation display field 1510 that displays equipment operations that should be performed by workers on a planned date, and an operation plan display field 1520.

In the operation display field 1510, the time 1511 at which each operation is performed, the content 1512 of each operation, and the operating conditions 1513 corresponding to each operation are displayed. Among the operations of each facility, stopping the pump 41 is displayed when the water flow rate (or the operating state of the pump 41) at a certain time is 0 and the water flow rate (or the operating state of the pump 41) in the previous time period is a positive number. Starting the pump 41 is displayed when the water flow rate (or the operating state of the pump 41) at a certain time is a positive number and the water flow rate (or the operating state of the pump 41) in the previous time period is 0. In addition, the operation display field 1510 displays "-" for derived operating conditions that were not extracted in s311 out of all derived operating conditions generated in the operating condition search process s3.

Details of the operation plan are displayed in the operation plan display field 1520. For example, the operation plan display field 1520 displays the state of each piece of equipment in each water purification plant 40 (flow rate of the pipeline 42 and pump 41, water level of the water reservoir 44, filtration flow rate of the pipeline 42, etc.) as each decision variable in the optimization model. The state of each piece of equipment may be graphed (trend graph, etc.) and displayed in the operation plan display field 1520.

As described above, the water purification plant operation planning device 10 of this embodiment acquires combinations (operation performance data 15) of the water state (such as the water level in the distribution reservoir 44) in each facility of the water purification plant 40 and the operation of each facility by the worker (such as the operation of the pump 41 and the flow rate of the pipeline 42), and calculates the operation conditions, which are the relationship between the water state in each facility of the water purification plant 40 and the operation of each facility by the worker corresponding to that state, based on the acquired combinations. Then, the water purification plant operation planning device 10 calculates the operation that the worker will perform on each facility in the water purification plant 40 on the planned date under the operating conditions, which best satisfies the constraints of each facility in the water purification plant 40.

That is, the water purification plant operation planning device 10 of the present embodiment sets the past state of each piece of equipment at the water purification plant 40 and each operation performed at that time as operation conditions, and calculates an operation plan (equipment operation, etc.) that satisfies the constraints of the water purification plant 40 based on these operation conditions. In this way, the water purification plant operation planning device 10 of the present embodiment can create an operation plan optimized for the water purification plant 40 that reflects, for example, operation operations that workers at the water purification plant 40 actually performed in the past. In this way, an operation plan can be created that accurately reflects the operation know-how of workers at the water purification plant.

The water purification plant operation planning device 10 of this embodiment also identifies similar past days that had state values for the water purification plant 40 on the planned date (e.g., the water level of the water reservoir 44) that were within a predetermined difference. Based on a combination of the water state and operation of each piece of equipment on these similar days, the water purification plant operation planning device 10 calculates the operating conditions by generating a trained model in which the water state of each piece of equipment in the water purification plant 40 is an explanatory variable and the operation of each piece of equipment in the water purification plant 40 is an objective variable.

In this way, by generating a trained model using data from equipment on similar days in the past when the equipment condition is similar to the planned date as training data and determining operating conditions, appropriate operating conditions can be generated that are in line with the condition on the planned date, even if the condition of each piece of equipment at the water purification plant 40 varies from day to day.

In addition, the water purification plant operation planning device 10 of this embodiment sets the range of water conditions in each piece of equipment at the water purification plant 40 (for example, the range of water levels in the water reservoir 44) as conditions, and displays a screen showing information on the results of the corresponding operations of each piece of equipment performed by workers (for example, operation of the pump 41), allowing workers to grasp an overview of the operations that have been performed (and should be performed) at the water purification plant 40.

The water purification plant operation planning device 10 of this embodiment also generates multiple candidate operating conditions reconstructed for the state of each facility by classifying each facility of the water purification plant 40 in terms of the water flow path (system 45) in the water purification plant 40, and identifies candidates that satisfy specified conditions as operating conditions (derived operating conditions) from among the generated candidates.

In this way, by generating multiple candidate operating conditions by reconstructing the operating conditions from the perspective of the system 45 of the water purification plant 40 and identifying the operating conditions from these, it is possible to generate an operating plan using appropriate operating conditions even if the water purification plant 40 has a complex flow path configuration.

Regarding the above-mentioned specified conditions, the water purification plant operation planning device 10 of this embodiment estimates the state (planned value) of each piece of equipment in the water purification plant 40 that best satisfies the constraints of each piece of equipment under each candidate operating condition. The water purification plant operation planning device 10 then identifies the candidate operating condition that has the smallest deviation from the state (actual value) of each piece of equipment in the water purification plant in the past as the operating condition. In this way, the most appropriate operating condition can be set by setting the candidate that is the planned value that has the smallest deviation from the past actual value as the operating condition for each planned value, which is the theoretical value calculated from each candidate operating condition.

The water purification plant operation planning device 10 of this embodiment also identifies the characteristics (time periods that are confidence intervals, etc.) of the time when the range of water conditions in the water purification plant 40 equipment (e.g., the range between the upper and lower limits of the water level of the water distribution reservoir 44) was less than a predetermined range, and identifies candidate operating conditions that have a small deviation from the water purification plant equipment conditions in this confidence interval as operating conditions.

Like a confidence interval, workers need to narrow down the operation content, and by comparing with a period during which the accuracy of equipment operation is considered high, highly similar operating conditions can be adopted, allowing for the generation of a highly accurate operation plan. This is particularly true for the operation of the water purification plant 40, where the operation of the water level of the water reservoir 44 is extremely important. By determining the operating conditions based on a confidence interval with a small fluctuation range for the water level of the water reservoir 44, a highly accurate operation plan can be generated.

In addition, in this embodiment, the constraints of a specific piece of equipment in the water purification plant 40 (for example, the water level of the water reservoir 44) include a hard constraint and a soft constraint that is less stringent than the hard constraint, and the water purification plant operation planning device 10 calculates the operation for each piece of equipment that satisfies the hard constraint of the specific equipment and has the smallest deviation from the soft constraint under the operating conditions on the planning date. In this way, by considering two constraints with different degrees of constraint, it is possible to generate a more flexible operation plan that is in line with the actual situation on site.

The water purification plant operation plan creation device 10 of this embodiment also displays the contents of the operation plan, for example, the operations that workers will perform on each piece of equipment. This allows workers to easily know the operation that should be performed on the planned date.

In addition, the water purification plant operation planning device 10 of this embodiment transmits instruction signals to realize the operation plan (equipment operation), and the monitoring control device 20 controls each piece of equipment based on these instruction signals. This allows automatic operation of the water purification plant 40 that accurately reflects the operating know-how of the workers at the water purification plant 40.

The present invention is not limited to the embodiments described above, and includes various modifications. The above-described embodiments have been described in detail to provide a better understanding of the present invention, and are not necessarily limited to those having all of the configurations described.

For example, some of the functions provided by each device in each embodiment may be provided in another device, and functions provided by another device may be provided in the same device.

In addition, in this embodiment, the operation of the equipment in the water purification plant 40 is the start or stop of the pump 41 or the flow rate of the pipeline 42, but it may be other operational operations such as adjusting the filtration flow rate of the water.

In addition, in this embodiment, upper and lower limits for the operating water level of the water reservoir 44 are set as soft constraints, but other constraints, such as the start-up sequence or start-up interval of the pumps 41, may also be set.

In addition, in this embodiment, the information on the operating conditions is generated by a trained model, but other information formats are also possible. For example, it may be a combination of multiple conditional expressions, or a database representing the operating conditions may be constructed.

In addition, in this embodiment, the method of selecting the derived operational conditions to be used in the optimization model from multiple derived operational conditions is to compare planned values from the operational plan with past actual values and select the derived operational conditions that are closest to the actual values, but other evaluation models or formulas that determine whether each derived operational condition is appropriate may also be used. Also, the user may be allowed to select appropriate derived operational conditions.

1 Water purification plant operation management system, 40 Water purification plant, 10 Water purification plant operation planning device, 14 Operation planning unit, 131 Operation condition extraction unit

Claims (10)

A processor and a memory,
an operational condition extraction unit that acquires a combination of the state of water in each facility of the water purification plant and the operation of each facility performed by an operator, and calculates an operational condition that is a relationship between the state of water in each facility of the water purification plant and the operation of each facility performed by an operator corresponding to the state, based on the acquired combination;
An operation planning unit that calculates an operation to be performed by an operator on each piece of equipment at the water purification plant at a predetermined time in the future under the operating conditions that best satisfies the constraints of each piece of equipment at the water purification plant;
A water purification plant operation planning device comprising: The operation condition extraction unit identifies a past time at which a state value related to the water purification plant at the predetermined future time point had a state value within a predetermined difference, and acquires a combination of the state of water in each piece of equipment of the water purification plant and the operation of each piece of equipment performed by an operator at the identified past time point,
and an operational condition learning unit that calculates operational conditions by generating a trained model based on the acquired combination, the trained model having the water state of each piece of equipment in the water purification plant as an explanatory variable and the operation of each piece of equipment in the water purification plant as a target variable.
The water purification plant operation planning device according to claim 1. The state of the water in the water purification plant under the operating conditions is expressed as a range of the state of the water in each of the facilities;
A display unit is further provided that displays information on the range of water conditions in each piece of equipment of the water purification plant based on the calculated operating conditions, and the results of the operation of each piece of equipment performed by an operator in response to the condition.
The water purification plant operation planning device according to claim 1. The water purification plant operation planning device according to claim 1 further comprises an operation condition search unit that generates multiple operation condition candidates reconfigured for different combinations of each piece of equipment by classifying each piece of equipment in the acquired combination with respect to the water flow path in the water purification plant, and identifies a candidate that satisfies a predetermined condition as the operation condition from each of the generated candidates. The operation condition search unit includes:
Generate candidates for the plurality of operating conditions based on the combination acquired by the operating condition extraction unit, and estimate a state of each piece of equipment in the water purification plant that best satisfies the constraints of the equipment under each of the generated candidates for the operating conditions;
acquiring a past status of each piece of equipment of the water purification plant, and identifying, as the operating condition, a candidate for the operating condition whose deviation from the acquired status of each piece of equipment is equal to or less than a predetermined value;
The water purification plant operation planning device according to claim 4. An evaluation time calculation unit that identifies characteristics of a time during which the range of the water state in the equipment of the water purification plant is less than a predetermined range,
The operation condition search unit acquires the state of the equipment of the water purification plant at a past time point having the specified characteristics, and identifies, as the operation condition, a candidate of the operation condition whose deviation from the acquired state of each equipment is equal to or less than a predetermined value.
The water purification plant operation planning device according to claim 5. The constraints of a given facility at the water purification plant include a first constraint and a second constraint that is less stringent than the first constraint;
the operation planning unit calculates an operation to be performed by the operator on each piece of equipment at the predetermined time point in the future under the operating conditions, which satisfies the first constraint of the predetermined piece of equipment and has a minimum deviation from the second constraint;
The water purification plant operation planning device according to claim 1. The water purification plant operation planning device according to claim 1, further comprising a display unit that displays the calculated operations to be performed by the worker on each piece of equipment. An information processing device,
an operation condition extraction process for acquiring a combination of the water state in each facility of the water purification plant and the operation of each facility performed by an operator, and calculating an operation condition, which is a relationship between the water state in each facility of the water purification plant and the operation of each facility performed by an operator in response to the water state, based on the acquired combination;
An operation plan creation process for calculating an operation to be performed by an operator on each piece of equipment at the water purification plant that best satisfies the constraints of each piece of equipment at a predetermined time in the future under the operating conditions;
A method for planning and developing water purification plant operations. A processor and a memory,
an operation condition extraction unit that acquires a combination of the water state in each facility of the water purification plant and the operation of each facility performed by an operator, and calculates an operation condition, which is a relationship between the water state in each facility of the water purification plant and the operation of each facility performed by an operator corresponding to the water state, based on the acquired combination;
An operation planning unit that calculates an operation to be performed on each piece of equipment in the water purification plant that best satisfies the constraints of each piece of equipment in the water purification plant under the operating conditions at a predetermined time in the future and transmits a signal instructing the calculated operation;
a monitoring and control device that receives the signal and controls each of the facilities based on the received signal;
A water purification plant operation management system equipped with the following:

Images