JP7596207B2 - Information processing device, data extraction method, machine learning model generation method, and data extraction program - Google Patents

Description

The present invention relates to an information processing device that processes various data measured in a plant.

In various industrial plants, various data are measured and utilized using various measuring devices. For example, the following Patent Document 1 discloses a technology for diagnosing abnormalities in a plant using plant data measured in the plant. More specifically, the technology of Patent Document 1 diagnoses abnormalities in the plant using a learning mechanism generated by machine learning using plant data.

Japanese Patent Application Publication No. 6-174503

Patent Document 1 does not disclose any method for appropriately selecting plant data to be used for learning from the wide variety of data measured in the plant. However, in order to perform abnormality diagnosis with high accuracy, it is extremely important to appropriately select the plant data to be used for learning. This is because, while the amount and variety of data measured in a typical plant is often enormous, if data that is less relevant or unrelated to plant abnormalities is used for learning, the accuracy of abnormality diagnosis will actually decrease.

For this reason, in the past, in order to use data measured at a plant, data users were required to acquire sufficient knowledge about the plant in advance. For example, in order to select relevant data, tasks such as examining the plant's process flow and equipment drawings have traditionally been performed. However, such tasks are time-consuming, and there is a problem in that the accuracy of data selection varies depending on the level of knowledge about the plant itself and the various equipment contained therein. Note that this type of problem is not limited to the selection of data to be used in machine learning, but is a common issue that arises when trying to use data measured at a plant.

One aspect of the present invention aims to realize an information processing device or the like that enables measurement data to be appropriately used even if the person does not have knowledge of the plant.

In order to solve the above problem, an information processing device according to one aspect of the present invention includes a position identification unit that identifies the measurement position of each piece of measurement data measured for each of a plurality of pieces of equipment arranged in a plant, and a related data extraction unit that extracts, as related data, pieces of the measurement data that are mutually related based on the measurement positions identified by the position identification unit.

In order to solve the above problem, an information processing device according to another aspect of the present invention includes a time difference determination unit that determines the difference in timing at which a specific target passes through each piece of equipment in a plant where multiple pieces of equipment are connected in a specific order, and a related data extraction unit that extracts, as related data, those pieces of time-series measurement data measured for each piece of equipment that are related to each other, based on the timing difference determined by the time difference determination unit.

A data extraction method according to one aspect of the present invention is a data extraction method executed by an information processing device, and includes a position identification step of identifying a measurement position of each piece of measurement data measured for each of a plurality of pieces of equipment arranged in a plant, and a related data extraction step of extracting, as related data, pieces of the measurement data that are mutually related based on the measurement positions identified in the position identification step.

In addition, a data extraction method according to another aspect of the present invention is a data extraction method executed by an information processing device, and includes a time difference identification step of identifying a difference in timing at which a predetermined object passes through each piece of equipment in a plant where the plurality of pieces of equipment are connected in a predetermined order, and a related data extraction step of extracting, as related data, pieces of the measurement data in a time series measured for each piece of equipment that are related to each other, based on the timing difference identified in the time difference identification step.

A method for generating a machine learning model according to one aspect of the present invention is a method for generating a machine learning model executed by one or more information processing devices, and includes a position identification step for identifying a measurement position of each piece of measurement data measured for each of a plurality of pieces of equipment arranged in a plant, a related data extraction step for extracting, as related data, pieces of measurement data that are mutually related based on the measurement positions identified in the position identification step, a teacher data generation step for generating teacher data for a machine learning model using the related data, and a learning step for generating the machine learning model by machine learning using the teacher data.

According to one aspect of the present invention, it becomes possible to appropriately use measurement data even if one does not have knowledge about the plant.

1 is a block diagram showing an example of a configuration of a main part of an information processing device according to a first embodiment of the present invention; FIG. 1 is a diagram showing an example of a waste incineration plant to which the information processing device can be applied. 11A and 11B are diagrams illustrating an example of measurement data recorded in a measurement data storage unit. FIG. 13 is a diagram showing an example in which the measurement position of a portable measuring device has changed. FIG. 1 is a diagram showing an example of the arrangement of structures and measuring instruments in a plant. FIG. 4 is a diagram illustrating an example of a process executed by the information processing device. FIG. 11 is a block diagram showing an example of a configuration of a main part of an information processing device according to a second embodiment of the present invention. FIG. 13 is a diagram illustrating an example of identifying a time difference and extracting related data. FIG. 4 is a diagram illustrating an example of a process executed by the information processing device.

[Embodiment 1]
(Configuration of information processing device 1)
The configuration of an information processing device 1 according to this embodiment will be described. The information processing device 1 is a device that processes measurement data measured for a plurality of equipment devices arranged in a plant. Here, a "plant" refers to a facility used industrially, and as described above, is equipped with a plurality of equipment devices, and these equipment devices perform predetermined processes, such as the production of products and the treatment of waste. An "equipment device" is a device that performs a predetermined process in a plant. One piece of equipment device may be a single device or may be made up of a plurality of devices.

Figure 1 is a block diagram showing an example of the main configuration of information processing device 1. As shown in the figure, information processing device 1 includes a control unit 11 that controls each unit of information processing device 1, and a storage unit 12 that stores various data used by information processing device 1. Information processing device 1 also includes a communication unit 13 that allows information processing device 1 to communicate with other devices, an input unit 14 that accepts input of various data to information processing device 1, and an output unit 15 that allows information processing device 1 to output various data.

The control unit 11 includes a measurement data acquisition unit 111, a position identification unit 112, a data management unit 113, a related data extraction unit 114, a teacher data generation unit 115, and a learning unit 116. The memory unit 12 includes a measurement data storage unit 121 and a model storage unit 122.

The measurement data acquisition unit 111 acquires measurement data measured for each of the multiple equipment devices arranged in the plant. The method of acquiring the measurement data is not particularly limited, and for example, a user of the information processing device 1 may input the measurement data via the input unit 14, in which case the measurement data acquisition unit 111 acquires the measurement data input via the input unit 14. Also, for example, the measurement data acquisition unit 111 may acquire the measurement data from a measuring device that measures the measurement data by communicating with the measuring device via the communication unit 13.

The position identification unit 112 identifies the measurement position of each piece of measurement data acquired by the measurement data acquisition unit 111. The method of identifying the measurement position is not particularly limited. For example, the position identification unit 112 may identify the position of the equipment corresponding to the measurement data, or the position of the measuring equipment that measured the measurement data, as the measurement position of the measurement data. In this case, the position of the equipment in the plant or the position of the measuring equipment in the plant is identified in advance. Then, the position identification unit 112 identifies the measurement position of the measurement data according to which equipment the measurement data was measured for, or which measuring equipment the measurement data was measured by. In addition, the position identification unit 112 may identify the measurement position by, for example, having the user input the measurement position of each piece of measurement data via the input unit 14.

The data management unit 113 associates the measurement data acquired by the measurement data acquisition unit 111 with location information indicating the measurement location of the measurement data identified by the location identification unit 112, and records them in the measurement data storage unit 121.

The related data extraction unit 114 extracts, as related data, measurement data that is mutually related among the measurement data based on the measurement positions identified by the position identification unit 112. It is only necessary to determine in advance what type of measurement data is to be treated as related data. For example, the related data extraction unit 114 may extract, as related data, measurement data measured at a measurement position within a predetermined range of distance to a target equipment device selected from among multiple equipment devices, or to a specified position specified within the plant.

Measurement data measured at a measurement position whose distance to the target equipment falls within a specified range is likely to be related to the target equipment. Similarly, measurement data measured at a measurement position whose distance to a designated position falls within a specified range is likely to be related to an event that occurred at the designated position. Therefore, with the above configuration, measurement data related to the target equipment or an event that occurred at the designated position can be extracted as related data.

The teacher data generation unit 115 generates teacher data for the machine learning model using the related data. By including the teacher data generation unit 115, the information processing device 1 can automatically generate teacher data for the machine learning model.

The machine learning model may be one that uses related data to make inferences about the plant. For example, the related data extraction unit 114 extracts measurement data measured near the target equipment or a specified location as related data. Note that the target equipment or the specified location may be specified by the user or may be determined in advance.

In this case, the teacher data generation unit 115 may generate teacher data for a machine learning model that performs a predetermined inference regarding the target equipment or the specified location. Examples of the predetermined inference include prediction of the operating state of the target equipment after a predetermined time, prediction of the probability of a predetermined event such as an operational abnormality occurring in the target equipment or the specified location, and prediction of the content of plant control required to prevent the occurrence of the event.

The teacher data generation unit 115 generates teacher data by associating the related data with correct answer data according to the content of the inference performed by the machine learning model. For example, assume that a machine learning model is generated that predicts whether or not an operational abnormality will occur in a target facility device within one hour. In this case, the teacher data generation unit 115 simply associates the related data with correct answer data that indicates whether or not an operational abnormality occurred within one hour after the related data was measured.

The learning unit 116 generates a machine learning model by machine learning using the teacher data generated by the teacher data generation unit 115. The machine learning algorithm may be appropriately selected depending on the content of the inference and the required accuracy of the inference. For example, a neural network or linear regression may be applied. The generated model is recorded in the model storage unit 122.

As described above, the measurement data storage unit 121 records the measurement data acquired by the measurement data acquisition unit 111 in association with location information indicating the measurement location of the measurement data. Details of the data recorded in the measurement data storage unit 121 will be described later with reference to FIG. 3. In addition, the model storage unit 122 records the machine learning model generated by the learning unit 116 as described above.

As described above, the information processing device 1 includes a position identification unit 112 that identifies the measurement position of each piece of measurement data measured for each of a plurality of pieces of equipment arranged in a plant, and a related data extraction unit 114 that extracts, as related data, pieces of the measurement data that are mutually related based on the measurement positions identified by the position identification unit 112.

There is often a correlation between the measurement data taken for each piece of equipment installed in a plant and the location where the data was measured. For example, when a fault occurs in a plant, the effects of the fault will often be apparent in measurement data taken at locations close to where the fault occurred, but not in measurement data taken at locations far from where the fault occurred.

Therefore, according to the above configuration in which related data is extracted based on the measurement position of the measurement data, it becomes possible to extract as related data those pieces of measurement data that are mutually related. And if the related data can be extracted, it becomes possible to use them appropriately even if one does not have knowledge about the plant.

As described above, the position identification unit 112 may identify the position of the facility equipment corresponding to the measurement data, or the position of the measuring equipment that measured the measurement data, as the measurement position of the measurement data. The information processing device 1 may also include a data management unit 113 that associates the measurement data with position information indicating the position identified by the position identification unit 112 and records them. With this configuration, the acquired measurement data and the position information of the measurement data are recorded in association with each other, so that the added value of position information can be given to the measurement data.

For example, if measurement data associated with location information is used, even a user with no knowledge of the plant can analyze the relevance of measurement data measured at physically nearby locations. In addition, by recording location information as three-dimensional location information, it is possible to display the measurement location of the measurement data in VR (Virtual Reality) or AR (Augmented Reality). This makes it possible to perform plant monitoring with a sense of realism (on-site feel) using VR, and to provide data to on-site maintenance personnel using AR. In addition, if a problem or accident occurs in the plant, it is possible to move a PTZ (Panorama Tilt Zoom) camera, a remotely controlled or autonomous drone, a robot, etc. to the vicinity of the site to collect information.

(Application example of information processing device 1)
An application example of the information processing device 1 will be described with reference to Fig. 2. Fig. 2 is a diagram showing an example of a waste incineration plant 1000 to which the information processing device 1 can be applied.

The waste incineration plant 1000 includes a waste pit 1001 that temporarily stores waste brought in by a waste collection truck P, and an incinerator 1002 that incinerates the waste in the waste pit 1001. The waste pit 1001 and the incinerator 1002 are connected by a hopper 1003 that supplies waste to the incinerator 1002, and the waste in the waste pit 1001 is sent through the hopper 1003 to the incinerator 1002 and incinerated.

The garbage pit 1001 is provided with a monitoring room 1004. A control device for operating the crane 1006 and the like are provided in the monitoring room 1004. The above-mentioned information processing device 1 may be disposed, for example, in this monitoring room 1004. Of course, the location of the information processing device 1 is arbitrary, and it may be disposed, for example, outside the garbage incineration plant 1000.

The bottom of the garbage pit 1001 is a garbage storage section 1005 that stores garbage brought in by the garbage collection truck P. In the example of FIG. 2, the garbage stored in the garbage storage section 1005 is indicated by "G." The garbage G in the garbage storage section 1005 is dumped into a hopper 1003 by a crane 1006 provided in the garbage pit 1001, and is then sent to the incinerator 1002.

The incinerator 1002 includes a fire grate 1007, a flue 1008, and an ash outlet 1009. Waste sent into the incinerator 1002 is incinerated while being transported by the fire grate 1007. Smoke and other materials generated by the incineration of the waste are released from the flue 1008 after harmful substances and other materials have been removed. In addition, incineration ash and other materials generated by the incineration of the waste are discharged from the ash outlet 1009.

The waste incineration plant 1000 having the above-mentioned configuration is provided with measuring devices 3a to 3g. More specifically, measuring devices 3a and 3b are provided in the waste pit 1001, measuring devices 3c to 3e are provided near the grate 1007, measuring device 3f is provided near the ceiling of the incinerator 1002, and measuring device 3g is provided in the flue 1008. In the following, when there is no need to distinguish between measuring devices 3a to 3g, they will simply be referred to as measuring device 3.

The measuring device 3 is a device that measures the equipment included in the waste incineration plant 1000 and outputs measurement data showing the measurement results. The measurement object may be anything that contributes to the operation of the waste incineration plant 1000, and the measurement method may be appropriately adopted according to the measurement object. For example, if the measuring device 3f is a thermometer, the temperature inside the furnace of the equipment, the incinerator 1002, may be used as the measurement data. Also, if the measuring device 3g is a sensor that detects the concentration of nitrogen oxides, the concentration of nitrogen oxides generated in the equipment, the incinerator 1002, may be used as the measurement data. In addition, if the measuring device 3b is a camera, image data showing the appearance of the waste inside the equipment, the hopper 1003, may be used as the measurement data. In addition, for example, a vibration meter or the like may be used as the measuring device 3.

The "measurement data" may also include data indicating the operating conditions of the equipment. In this case, the equipment that controls the operation of the equipment may be the measuring equipment 3. For example, the equipment that controls the operation of the grate 1007 may be the measuring equipment 3, and the setting value of the grate speed may be the measurement data.

(Example of measurement data)
The measurement data recorded in the measurement data storage unit 121 will be described with reference to Fig. 3. Fig. 3 is a diagram showing an example of the measurement data recorded in the measurement data storage unit 121. More specifically, Fig. 3 shows data in a table format in which a measurement device ID, an equipment device ID, and location information are associated with each piece of measurement data.

The measurement device ID is identification information that indicates the measurement device 3 that output the measurement data. The facility device ID is identification information that indicates the facility device from which the measurement data was measured. And the location information is information that indicates the measurement location of the measurement data.

For example, in the example of FIG. 3, the measurement data (a0, a1, a2, ...) is time-series data, the measurement device ID is 0001, the facility device ID is A, and the location information is (x1, y1, z1). As in this example, the location information may be, for example, a representation of a position within the waste incineration plant 1000 using three-dimensional coordinate values.

From this information, it is possible to identify which measuring device 3 measured which facility equipment the measurement data was taken from, and also to identify the measurement location of the measurement data.

The measurement data (d0, d1, d2, ...) shown in Figure 3 has a measurement device ID of M004. It is assumed that the measurement device 3 indicated by this measurement device ID is a portable measurement device 3. "Portable" means that the measurement location within the plant can be changed. Since the measurement location of a portable measurement device 3 can be changed, for measurement data measured by a portable measurement device 3, it is necessary to identify the location of the measurement device 3 at the time of measurement. The portable measurement device 3 is transported to the measurement location by, for example, an on-site maintenance worker, a drone, or a robot.

(Handling of measurement data obtained using portable measuring equipment)
As described above, the measurement position of the portable measuring device 3 changes, so the position of the measuring device 3 when the measurement is performed must be regarded as the measurement position. In addition, as the measurement position changes, the associated data of the measurement data measured by the portable measuring device 3 may also change. This will be described with reference to Fig. 4. Fig. 4 is a diagram showing an example in which the measurement position of the portable measuring device 3 has changed.

In the example of FIG. 4, maintenance worker X at the plant carries portable measuring device 3B. At time t1, maintenance worker X performs a measurement using measuring device 3B near equipment A where measuring device 3A is installed. At time t2, maintenance worker X performs a measurement using measuring device 3B near equipment C where measuring device 3C is installed.

In this way, when measurements are made using portable measuring device 3B, the measurement data measured at time t1 and the measurement data measured at time t2 are different, even if the data are measured using the same measuring device 3B. In other words, the measurement position of the measurement data measured by measuring device 3B may change dynamically.

Therefore, for measurement data measured by the portable measuring device 3B, the position identification unit 112 identifies the position of the measuring device 3B at the time the measurement was performed as the measurement position. This makes it possible to identify a valid measurement position for measurement data measured by the portable measuring device 3B.

The method of identifying the measurement position of the portable measuring device 3B is not particularly limited. For example, when maintenance worker X performs a measurement with the measuring device 3B, the measurement position may be recorded and the measurement position may be input to the information processing device 1. Also, for example, beacon signal transmitters may be attached to various locations within the plant, and the receivers may be attached to the maintenance worker X or the measuring device 3B. In this case, the position identification unit 112 can identify the measurement position based on which location's beacon signal is received by the receiver during measurement of the measurement data. In addition, it is also possible to attach a GPS (Global Positioning System) receiver to the maintenance worker X or the measuring device 3B, and identify the measurement position using the position information output by the receiver.

The measurement data measured by the portable measuring device 3B can be extracted as related data for different measurement data depending on the measurement time. For example, in the example of FIG. 4, the measurement position of the measuring device 3B at time t1 is near the equipment A. Therefore, the measurement data measured by the measuring device 3B at time t1 can be extracted as related data for the measurement data measured by the measuring device 3A provided in the equipment A. On the other hand, the measurement position of the measuring device 3B at time t2 is near the equipment C, so the measurement data measured by the measuring device 3B at time t2 can be extracted as related data for the measurement data measured by the measuring device 3C provided in the equipment C.

(Example of extracting related data using structure information)
The related data extraction unit 114 may extract related data based on structure information indicating the arrangement of structures arranged in the plant. This will be described with reference to Fig. 5. Fig. 5 is a diagram showing an example of the arrangement of a structure ST and measuring instruments 3A to 3C in a plant. Fig. 5 shows the structure ST, which is an L-shaped wall when viewed from above, as viewed from directly above.

Measuring devices 3A to 3C are placed around this structure ST. The distance between measuring devices 3A and 3B is d1, and the distance between measuring devices 3B and 3C is d2. In the example of Figure 5, d2 > d1, so when extracting associated data related to the measurement data of measuring device 3B simply from the distance alone, it is possible that the measurement data of measuring device 3A will be extracted and the measurement data of measuring device 3C will not be extracted.

However, a structure ST exists between measuring devices 3A and 3B, and the presence of this structure ST may weaken or eliminate the correlation between the measurement data measured by measuring device 3A and the measurement data measured by measuring device 3B. For example, if measuring devices 3A to 3C are all temperature sensors and structure ST is a heat insulating wall that does not easily transmit heat, the correlation between the measurement data measured by measuring device 3A and the measurement data measured by measuring device 3B is likely to be weak. Therefore, in this case, it is more appropriate to extract the measurement data of measuring device 3C as the associated data related to the measurement data of measuring device 3B, rather than the measurement data of measuring device 3A.

In view of the above, the related data extraction unit 114 may extract related data based not only on the measurement positions identified by the position identification unit 112 but also on structure information indicating the placement of structures placed in the plant. With this configuration, related data is extracted based on the measurement positions and on the structure information indicating the placement of structures placed in the plant, making it possible to extract related data taking into account the placement of structures.

The method of extracting related data based on structure information may be any method that reflects the influence of the structure in the extraction results. For example, as shown in FIG. 5, the related data extraction unit 114 may set a route that goes around the structure ST to connect the measurement positions, and consider the path length of the route (d3 in the example of FIG. 5) to be the distance between the measurement positions of measuring device 3A and measuring device 3B. This is an effective method for measuring sound, vibration, and the like that travel around a structure.

Also, for example, when a structure exists between the measurement positions, such as measuring device 3A and measuring device 3B in FIG. 5, the related data extraction unit 114 may be configured not to extract the measurement data as related data.

In addition, for example, when a structure is present between the measurement positions, the related data extraction unit 114 may weight the distance between those measurement positions before extracting the related data. For example, in the example of FIG. 5, if the distance d1 between the measurement positions of the measuring device 3A and the measuring device 3B is multiplied by a weight of 1.5, the weighted d1 becomes larger than d2. Therefore, in this case, if one piece of measurement data that is most related to the measurement data of the measuring device 3B is extracted, the measurement data of the measuring device 3C is extracted, not the measurement data of the measuring device 3A. The weight value may be determined according to the type of measurement data, the shape, material, size, etc. of the structure ST.

The structure information may indicate the position and range of the structure arranged in the plant using coordinates or the like, or may indicate the room allocation in the plant. When using structure information indicating the room allocation in the plant, the related data extraction unit 114 may preferentially extract measurement data whose measurement positions are in the same room as related data.

(Processing flow)
The flow of the process executed by the information processing device 1 will be described with reference to Fig. 6. Fig. 6 is a diagram showing an example of the process executed by the information processing device 1. Among the processes in Fig. 6, steps S11 to S14 correspond to a data extraction method, and the entire process in Fig. 6 including steps S15 and S16 corresponds to a machine learning model generation method.

In S11, the measurement data acquisition unit 111 acquires the measurement data measured by each measuring device 3 during a predetermined period. The measurement data acquisition unit 111 may acquire measurement data received via the communication unit 13, or may acquire measurement data input via the input unit 14. Note that the method and timing of acquiring the measurement data are not limited to this example. For example, the measurement data acquisition unit 111 may sequentially acquire the measurement data output by each measuring device 3.

In S12 (location identification step), the location identification unit 112 identifies the measurement location of the measurement data acquired in S11, i.e., each piece of measurement data measured for each of a number of pieces of equipment installed in a plant (e.g., a waste incineration plant 1000).

For example, the position identification unit 112 may identify the position of the measuring device 3 as the measurement position of the measurement data. Also, for example, the position identification unit 112 may identify the position of the facility device from which the measurement data was measured as the measurement position of the measurement data. Also, if the measurement data acquired in S11 includes data measured by a portable measuring device 3, the position identification unit 112 may identify the measurement position by the method described based on FIG. 4.

In S13, the data management unit 113 associates the measurement data acquired in S11 with location information indicating the measurement location identified in S12 and records the data in the measurement data storage unit 121. As in the example of FIG. 3, the data management unit 113 may also record the identification information of the measuring device 3 and the identification information of the facility device.

In S14 (related data extraction step), the related data extraction unit 114 extracts, as related data, the measurement data acquired in S11 that are mutually related based on the measurement position identified in S12. For example, the related data extraction unit 114 may extract measurement data measured at a measurement position where the distance to the equipment specified by the user of the information processing device 1, i.e., the target equipment described above, is less than or equal to a predetermined value, as related data related to the target equipment.

In S15, the teacher data generation unit 115 generates teacher data by associating the correct answer data with the related data extracted in S14. Then, in S16, the learning unit 116 performs machine learning using the teacher data generated in S15, generates a machine learning model, and records the generated machine learning model in the model storage unit 122. This ends the processing in FIG. 6.

As described above, the data extraction method executed by the information processing device 1 includes a position identification step (S12) for identifying the measurement position of each piece of measurement data measured for each of a plurality of pieces of equipment arranged in the plant, and a related data extraction step (S14) for extracting, as related data, pieces of measurement data that are mutually related based on the measurement positions identified in the position identification step. This makes it possible to appropriately use the measurement data even if one does not have knowledge about the plant.

As described above, the method for generating a machine learning model executed by the information processing device 1 includes a position identification step (S12) for identifying the measurement position of each piece of measurement data measured for each of a plurality of pieces of equipment arranged in the plant, a related data extraction step (S14) for extracting, as related data, pieces of measurement data that are mutually related based on the measurement positions identified in the position identification step, a teacher data generation step (S15) for generating teacher data for the machine learning model using the related data, and a learning step (S16) for generating a machine learning model by machine learning using the teacher data. According to this generation method, it is possible to extract mutually related measurement data and generate a machine learning model even if one has no knowledge of the plant.

The method for generating the machine learning model may be executed by one information processing device 1 as in the above example, or may be executed by multiple information processing devices. In the latter case, for example, the method for generating the machine learning model may be executed using an information processing device 1 that performs the processes of S11 to S14 and an information processing device that performs the processes of S15 and S16. Also, the process of S15 and the process of S16 may be executed by different information processing devices.

[Embodiment 2]
Other embodiments of the present invention will be described below. For ease of explanation, the same reference numerals are given to members having the same functions as those described in the above embodiment, and the description thereof will not be repeated.

(Configuration of information processing device 1A)
7 is a block diagram showing an example of a main configuration of an information processing device 1A according to this embodiment. The information processing device 1A differs from the information processing device 1 in that the information processing device 1A includes a time difference identification unit 112A instead of the position identification unit 112, and in that the information processing device 1A includes a travel condition identification unit 117A.

The time difference identification unit 112A identifies the difference in timing when a specific target passes through each piece of equipment in the plant. Details of the processing performed by the time difference identification unit 112A will be described later with reference to FIG. 8. Then, the related data extraction unit 114 of the information processing device 1A extracts, as related data, pieces of time-series measurement data measured for each piece of equipment that are mutually related, based on the timing difference identified by the time difference identification unit 112A.

The movement condition identification unit 117A identifies the movement conditions of a specific target between equipment devices. The movement condition identification unit 117A is not a required component. When the movement condition identification unit 117A is provided, the time difference identification unit 112A identifies the difference in timing at which the specific target passes through each equipment device based on the movement conditions identified by the movement condition identification unit 117A. This makes it possible to appropriately identify the difference in timing at which the target passes through each equipment device even if the movement conditions of the target fluctuate. The method of identifying the movement conditions will be described later with reference to FIG. 8.

As described above, the information processing device 1A includes a time difference determination unit 112A that determines the difference in timing when a specific target passes through each piece of equipment in the plant, and a related data extraction unit 114 that extracts, as related data, time-series measurement data measured for each piece of equipment that are related to each other based on the timing difference determined by the time difference determination unit 112A.

In a plant where a specific object passes through multiple pieces of equipment connected in a specific order, there is a difference in the timing at which the object passes through each piece of equipment. Measurement data measured on equipment when the object is passing through the equipment is likely to be related to the object, while measurement data measured on equipment when the object is not passing through the equipment is likely to be unrelated to the object. Therefore, with the above configuration that extracts related data based on the difference in the timing at which the object passes through each piece of equipment, it becomes possible to extract related data from time-series measurement data.

(Example of identifying time zones and extracting related data)
The identification of a time difference by the travel condition identification unit 117A and the time difference identification unit 112A, and the extraction of related data by the related data extraction unit 114 based on the identified time difference will be described with reference to Fig. 8. Fig. 8 is a diagram showing an example of identifying a time difference and extracting related data.

In the example of Figure 8, it is assumed that the target plant is equipped with three pieces of equipment, A to C, and that equipment A to C are equipped with measuring devices 3A to 3C, respectively. Equipment A to C are connected in the order of A to C, and a specific target Y passes through equipment A to C in this order of connections. Target Y may be, for example, a gas (e.g., exhaust gas, etc.) passing through the inside or surroundings of each piece of equipment A to C, a liquid (e.g., processing liquid or cooling water being processed in the plant), or a solid.

In this example, the target plant is a waste incineration plant, and the target Y is waste. The target Y is transported from equipment A to equipment B by a first transport device (not shown), and is further transported from equipment B to equipment C by a second transport device (not shown).

In the example of Figure 8, at time t0, object Y is inside equipment A, and the value of the measurement data measured by measuring device 3A at this time is a0. At time t1, which is Δt1 after time t0, object Y has moved to equipment B, and the value of the measurement data measured by measuring device 3B at this time is b1. Then, at time t2, which is Δt2 after time t1, object Y has moved to equipment C, and the value of the measurement data measured by measuring device 3C at this time is c2.

In this example, the time difference determination unit 112A determines the difference between the timing at which the object Y passes through equipment A and the timing at which it passes through equipment B as Δt1, and determines the difference between the timing at which the object Y passes through equipment B and the timing at which it passes through equipment C as Δt2.

Various methods can be applied to identify such timing differences. For example, the movement condition identification unit 117A may be configured to identify the transport speed of the target Y between the equipment. In this case, the time difference identification unit 112A can calculate the timing difference from the identified transport speed and the distance between the equipment.

Also, for example, the standard movement conditions and standard movement time of target Y between equipment devices may be determined in advance. In this case, the time difference determination unit 112A may calculate the timing difference by correcting the standard movement time based on the difference between the movement conditions identified by the movement condition determination unit 117A and the standard movement conditions. For example, if the movement conditions identified by the movement condition determination unit 117A reduce the movement speed of target Y by 10% compared to the standard movement conditions, the time difference determination unit 112A may determine the timing difference as a time that is 10% longer than the standard movement time.

The movement condition identified by the movement condition identification unit 117A may be any condition that affects the difference in timing at which the specified target passes through each piece of equipment. For example, if the specified target is a liquid, the movement condition identification unit 117A may identify the flow rate or flow speed as the movement condition. Also, for example, if the specified target is a gas, the movement condition identification unit 117A may identify the wind speed as the movement condition.

The movement condition identification unit 117A may also identify the operating status of the transport device as the movement condition. Furthermore, if dampers, valves, etc. are provided between the equipment, the movement condition identification unit 117A may identify their operating state or open/closed state as the movement condition. By using such movement conditions, it becomes possible to extract related data that takes into account temporary stagnation of the target, changes in the movement route, etc.

The data management unit 113 may record the timing difference identified in the above manner as time difference information in the measurement data storage unit 121. FIG. 8 shows an example of measurement data recorded in the measurement data storage unit 121 in association with time difference information. By recording the measurement data and the time difference information in association with each other in this manner, the related data extraction unit 114 can extract related data by referring to the measurement data storage unit 121.

More specifically, each measurement data shown in FIG. 8 is associated with a measurement device ID, an equipment ID, time difference information, and information indicating the measurement time. As described above, the time difference information indicates the difference in timing when a specified target passes through the equipment. The measurement time indicates the measurement time of each measurement data.

The related data extraction unit 114 can extract related measurement data as related data by referring to the measurement data associated with such various information. For example, assume that the related data of a0 (measurement time is t0) surrounded by a dashed line is extracted from the measurement data measured for the equipment A shown in FIG. 8. In this case, the related data extraction unit 114 extracts measurement data b1 measured at a time (t0+Δt1=t1) when the time Δt1 indicated by the time difference information has elapsed from the time t0 when a0 was measured from the measurement data measured for the equipment B located downstream of the equipment A. In addition, the related data extraction unit 114 extracts measurement data c2 measured at a time (t1+Δt2=t2) when the time Δt2 indicated by the time difference information has elapsed from the time t1 when b1 was measured from the measurement data measured for the equipment C located downstream of the equipment B. Since each of these measurement data is considered to reflect the influence of the presence of the object Y, each of these measurement data can be said to be related data related to the object Y.

(Processing flow)
The flow of the process executed by the information processing device 1A will be described with reference to Fig. 9. Fig. 9 is a diagram showing an example of the process executed by the information processing device 1A. Of the processes in Fig. 9, S21 to S25 correspond to a data extraction method, and the entire process in Fig. 9 including S26 and S27 corresponds to a machine learning model generation method.

In S21, similar to S11 in FIG. 6, the measurement data acquisition unit 111 acquires measurement data. In addition, in S22, the movement condition identification unit 117A identifies the movement conditions under which a specified target moves between equipment devices of the plant during the measurement period of the measurement data acquired in S21.

In S23 (time difference determination step), the time difference determination unit 112A determines the difference in timing when a predetermined target passes each piece of equipment based on the movement conditions determined in S22. Then, in S24, the data management unit 113 associates the time difference determined in S23 with the measurement data acquired in S21 and records it in the measurement data storage unit 121.

In S25 (related data extraction step), the related data extraction unit 114 extracts, as related data, data that is related to each other from the time-series measurement data measured for each piece of equipment based on the timing difference identified in S23. The processes in S26 and S27 are similar to S15 and S16 in FIG. 6.

As described above, the data extraction method executed by the information processing device 1A includes a time difference identification step (S23) of identifying the difference in timing at which a specific target passes through each piece of equipment in a plant where the specific target passes through multiple pieces of equipment connected in a specific order, and a related data extraction step (S25) of extracting, as related data, pieces of time-series measurement data measured for each piece of equipment that are related to each other based on the timing difference identified in the time difference identification step. With this configuration, it becomes possible to extract pieces of time-series measurement data that are related to each other as related data.

The method for generating a machine learning model executed by the information processing device 1A includes a time difference identification step (S23) for identifying the difference in timing at which a predetermined target passes through each piece of equipment in a plant where the target passes through multiple pieces of equipment connected in a predetermined order in the order of the connections, a related data extraction step (S25) for extracting, as related data, time-series measurement data measured for each piece of equipment that are related to each other based on the timing difference identified in the time difference identification step, a teacher data generation step (S26) for generating teacher data for a machine learning model using the related data, and a learning step (S27) for generating the machine learning model by machine learning using the teacher data. According to this generation method, it is possible to extract related measurement data and generate a machine learning model even if one does not have knowledge about the plant.

[Modifications]
The control unit 11 of the information processing device 1A of the second embodiment may be added with the position identification unit 112 included in the information processing device 1 of the first embodiment. In this case, the related data extraction unit 114 may extract related data from the time-series measurement data measured for each of the facility devices based on the measurement position identified by the position identification unit 112 and the timing difference identified by the time difference identification unit 112A.

The above configuration extracts related data based on not only the difference in timing when the target passes through each piece of equipment, but also the measurement position, making it possible to extract related data from the time-series measurement data.

The related data extracted in the first and second embodiments can be used as training data as described above, and can also be used for various other purposes. For example, the extracted related data can be analyzed to predict the state of the plant, and the cause of operational abnormalities that occur in the plant can be analyzed.

The time difference information described in the second embodiment can also be used as tracking information for the target. For example, in a plant that manufactures products, the time difference information can be used to extract measurement data measured by each piece of equipment during the manufacture of each product as related data, and these can be used as quality control data for each product.

Time difference information can also be used to correct measurement data. For example, a fault in one piece of equipment can cause measurement data from downstream equipment to become abnormal. In such cases, the time difference information can be used to identify the measurement data that has become abnormal due to the fault, and it can also be used to correct such measurement data to eliminate the effects of the fault.

[Software implementation example]
The functions of information processing device 1 and information processing device 1A (hereinafter referred to as the "device") can be realized by a program for causing a computer to function as the device, and a program (data extraction program) for causing a computer to function as each control block of the device (particularly each part included in the control unit 11).

In this case, the device includes a computer having at least one control device (e.g., a processor) and at least one storage device (e.g., a memory) as hardware for executing the program. The control device and storage device execute the program, thereby realizing each of the functions described in each of the above embodiments.

The program may be recorded on one or more computer-readable recording media, not on a temporary basis. The recording media may or may not be included in the device. In the latter case, the program may be provided to the device via any wired or wireless transmission medium.

In addition, some or all of the functions of each of the above control blocks can be realized by a logic circuit. For example, an integrated circuit in which a logic circuit that functions as each of the above control blocks is formed is also included in the scope of the present invention. In addition, it is also possible to realize the functions of each of the above control blocks by, for example, a quantum computer.

The present invention is not limited to the above-described embodiments, and various modifications are possible within the scope of the claims. The technical scope of the present invention also includes embodiments obtained by appropriately combining the technical means disclosed in different embodiments.

1, 1A Information processing device 112 Position identification unit 112A Time difference identification unit 113 Data management unit 114 Related data extraction unit 115 Teacher data generation unit 116 Learning unit 117A Travel condition identification unit

Claims (9)

a position identifying unit that identifies a measurement position of each piece of measurement data measured for each of a plurality of equipment devices arranged in the plant;
a related data extraction unit that extracts, as related data, data that is mutually related among the measurement data based on the measurement position identified by the position identification unit ,
The information processing device, wherein the related data extraction unit extracts the related data based on structure information indicating the position and range of a structure placed in the plant or room allocation in the plant and the measurement position identified by the position identification unit . The position identification unit identifies a position of the facility equipment corresponding to the measurement data or a position of a measuring equipment that measured the measurement data as a measurement position of the measurement data;
The information processing apparatus according to claim 1 , further comprising a data management unit configured to record the measurement data in association with position information indicating the position identified by the position identification unit. The information processing device according to claim 1 or 2, wherein the related data extraction unit extracts, as the related data, the measurement data measured at the measurement position where the distance to a target equipment device selected from among the plurality of equipment devices or the distance to a designated position designated within the plant is within a predetermined range. The information processing device according to any one of claims 1 to 3, further comprising a teacher data generation unit that generates teacher data for a machine learning model using the associated data. The information processing device according to claim 1 , wherein the position identifying unit identifies, for the measurement data measured by a portable measuring device, a position of the measuring device when the measurement was performed as the measurement position. a time difference specification unit that calculates a difference in timing when a predetermined object passes through each piece of equipment from a transport speed at which the predetermined object is transported between the plurality of equipment devices connected in a predetermined order and the distance between the equipment devices, or calculates the difference in timing by correcting a standard travel time of the object between the plurality of equipment devices based on a difference between a standard travel condition of the object between the plurality of equipment devices and an actual travel condition of the object;
6. The information processing device according to claim 1, wherein the related data extraction unit extracts the related data from the time-series measurement data measured for each of the facility devices based on the measurement position identified by the position identification unit and the timing difference calculated by the time difference identification unit . A data extraction method executed by an information processing device, comprising:
a position specifying step of specifying a measurement position of each piece of measurement data measured for each of a plurality of equipment devices arranged in the plant;
a related data extraction step of extracting, as related data, data that is mutually related among the measurement data based on the measurement positions identified in the position identification step ;
A data extraction method in which, in the relevant data extraction step, the relevant data is extracted based on structure information indicating the position and range of a structure placed in the plant or room allocation in the plant, and the measurement position identified in the position identification step . A method for generating a machine learning model executed by one or more information processing devices, comprising:
a position specifying step of specifying a measurement position of each piece of measurement data measured for each of a plurality of equipment devices arranged in the plant;
a related data extraction step of extracting, as related data, data that is mutually related among the measurement data based on the measurement positions identified in the position identification step;
A training data generation step of generating training data for a machine learning model using the related data;
A learning step of generating the machine learning model by machine learning using the teacher data ,
A method for generating a machine learning model, in which the relevant data extraction step extracts the relevant data based on structure information indicating the position and range of structures placed in the plant, or room allocation in the plant, and the measurement position identified in the position identification step . A data extraction program for causing a computer to function as the information processing device according to claim 1, the data extraction program causing a computer to function as the position identification unit and the related data extraction unit.

Images