Water treatment control support system, water treatment control support method, and water treatment control support program

Abstract

To obtain a system for supporting water-treatment control capable of carrying out operation support matched with an operation condition.SOLUTION: A system for supporting water-treatment control 2 includes an inference apparatus 20 that infers an operation candidate 53 which is a candidate for an operation in a case where an alarm is generated in a water treatment plant 1. The inference apparatus 20 comprises: an inference-information acquiring section 21 that acquires, as for-inference data, generated-alarm information 63B indicative of a type of the alarm and an operation condition 61B indicative of a condition having an effect upon an operation for responding to the alarm; and an inference section 22 that infers the operation candidate 53 by inputting the for-inference data to a learnt model 50 for inferring the operation candidate 53 from the for-inference data and outputs the inferred operation candidate 53.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] The present disclosure relates to a water treatment control support system, a water treatment control support method, and a water treatment control support program for assisting the operation of water treatment control in a water treatment plant.

Background Art

[0002] In a water treatment plant such as a water supply and sewerage plant, an operator monitors and controls the water treatment plant using a monitoring control device. When an alarm occurs in the water treatment plant, a control operation corresponding to the alarm is performed at the discretion of the operator.

[0003] For example, the plant operation support device described in Patent Document 1 registers an operation setting screen corresponding to an alarm in a screen registration database, and when an alarm occurs, supports the operation of the plant by displaying the operation setting screen corresponding to the alarm.

Prior Art Documents

Patent Documents

[0004]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0005] However, in the technology of Patent Document 1 above, when an operation corresponding to an alarm is actually executed, an operation corresponding to the operation condition is required. Nevertheless, even when the operation conditions are different, the same operation setting screen corresponding to the alarm is displayed. Therefore, with the technology of Patent Document 1 above, it has been impossible to provide operation support according to the operation conditions to the operator.

[0006] The present disclosure has been made in view of the above, and an object thereof is to obtain a water treatment control support system capable of providing operation support according to operation conditions. [Means for solving the problem]

[0007] To solve the above-mentioned problems and achieve the objectives, the water treatment control support system disclosed herein has an inference device that infers candidate operations, which are candidates for operations when an alarm occurs in a water treatment plant. The inference device comprises an inference information acquisition unit that acquires alarm occurrence information indicating the type of alarm and operation conditions indicating conditions that affect operations to deal with the alarm as inference data, and an inference unit that infers candidate operations corresponding to the alarm occurrence information and operation conditions from the alarm occurrence information and operation conditions by inputting the inference data into a trained model for inferring candidate operations from the inference data, and outputs the inferred candidate operations. The operating conditions include external factors when responding to an alarm, and these external factors include at least one of the following when responding to an alarm: season, day of the week, time of day, and weather. [Effects of the Invention]

[0008] The water treatment control support system described herein has the effect of providing operational support according to operating conditions. [Brief explanation of the drawing]

[0009] [Figure 1] This diagram shows the configuration of a water treatment plant equipped with a water treatment control support system according to the embodiment. [Figure 2] A diagram illustrating an example of the learning phase in the water treatment control support system according to the embodiment. [Figure 3] A diagram illustrating the neural network used in the learning device according to the embodiment. [Figure 4] A flowchart illustrating the processing procedure of the learning process performed by the learning device according to the embodiment. [Figure 5] A diagram illustrating an example of the utilization phase in the water treatment control support system according to the embodiment. [Figure 6] A flowchart showing the processing steps of the inference process performed by the inference device according to the embodiment. [Figure 7]This figure shows an example of the configuration of a processing circuit when the processing circuit of the water treatment control support system according to the embodiment is implemented using a processor and memory. [Figure 8] This figure shows an example of a processing circuit when the processing circuit of the water treatment control support system according to the embodiment is configured with dedicated hardware. [Modes for carrying out the invention]

[0010] The water treatment control support system, water treatment control support method, and water treatment control support program according to embodiments of this disclosure will be described in detail below with reference to the drawings.

[0011] Embodiment. Figure 1 shows the configuration of a water treatment plant equipped with a water treatment control support system according to an embodiment. The water treatment plant 1 comprises a water treatment control support system 2, a monitoring and control device 30, a display device 40, and a water treatment device (not shown) which will be described later.

[0012] The water treatment control support system 2 is a system that supports the operation of water treatment control in the water treatment plant 1. The water treatment control support system 2 provides operation support (control assistance) to the operator of the water treatment plant 1 by providing operation candidates 53 when an alarm occurs. Operation candidates 53 are candidates for the actions to be taken to deal with the alarm that has occurred. Operation candidates 53 include the type of display screen that the monitoring and control device 30 displays on the display device 40 (graphic screen, parameter adjustment screen, etc.), and control operations on the display screen (stop command to equipment, change of flow rate setting value, etc.).

[0013] The water treatment plant 1 of the present embodiment generates a learned model 50 for inferring an optimal operation candidate 53 corresponding to an alarm that has occurred by performing learning using operation result data 51A indicating past operation results to the monitoring control device 30 as teacher data. Then, the water treatment plant 1 uses the generated learned model 50 to infer an optimal operation candidate 53 corresponding to the alarm in progress, and displays the operation candidate 53 on the screen of the display device 40 connected to the monitoring control device 30. As a result, the operator of the monitoring control device 30 can efficiently and quickly respond to alarms.

[0014] The water treatment plant 1 is a plant that is arranged in at least one of a water supply system and a sewage system and performs water treatment. The water treatment plant 1 has various devices for performing water treatment (hereinafter referred to as water treatment devices), and an operator controls the water treatment by sending commands to these water treatment devices. When an alarm occurs, the operator executes an operation for water treatment control with respect to the monitoring control device 30 by referring to the operation candidate 53 provided from the water treatment control support system 2.

[0015] The monitoring control device 30 is a device that monitors the water treatment plant 1 and controls the water treatment plant 1 according to instructions from an operator. For example, the monitoring control device 30 monitors the water treatment devices and controls the water treatment devices according to instructions from an operator. When the monitoring control device 30 receives the operation candidate 53 sent from the water treatment control support system 2, it causes the display device 40 to display the operation candidate 53. The display device 40 is a display device such as a liquid crystal monitor that displays the operation candidate 53.

[0016] The water treatment control support system 2 includes a learning device 10, a storage device 60, and an inference device 20. The learning device 10, the storage device 60, and the inference device 20 may be connected by a network or the like, or may be realized as one computer. The learning device 10 is a computer that generates a learned model 50 for inferring an operation candidate 53 that is a candidate for an optimal next operation from operation result data 51A indicating past operation results to the monitoring control device 30.

[0017] The operation performance data 51A includes an operation condition (operation condition 61A described later) indicating a condition that affects an operation for dealing with an alarm, alarm information (generated alarm information 63A described later) which is information on the generated alarm, and operation content (operation content 62A described later) which is information indicating the content of the operation executed for the alarm. The operation condition 61A includes information on the situation when the alarm occurs and information on the operation for the alarm. The learning device 10 stores the learned model 50 in the storage device 60. The storage device 60 stores the learned model 50 generated by the learning device 10.

[0018] The inference device 20 is a computer that uses the learned model 50 generated by the learning device 10 and stored in the storage device 60 to infer an optimal operation candidate 53 corresponding to the ongoing alarm.

[0019] Note that the storage device 60 may be arranged inside the learning device 10, or may be arranged inside the inference device 20. Also, the storage device 60 may be arranged outside the water treatment control support system 2.

[0020] Also, in FIG. 1, the case where the learning device 10 and the inference device 20 are realized by different computers is described, but the learning device 10 and the inference device 20 may be realized by one computer. Also, at least one of the learning device 10 and the inference device 20 may exist on a cloud server.

[0021] The learning device 10 comprises a data acquisition unit 11 and a learning model generation unit 12. The data acquisition unit 11 acquires multiple operation record data 51A that show operation records. Operation record data 51A is operation record data before the water treatment control support system 2 was introduced. The data acquisition unit 11 also acquires operation record data 51B (not shown), which will be described later, as operation record data after the water treatment control support system 2 was introduced. Operation record data 51B is operation record data when operation was performed while referring to operation candidate 53. Operation record data 51A and 51B are information on the operation history of the operator to the monitoring control device 30 when an alarm occurs.

[0022] Before the introduction of the Water Treatment Control Support System 2, when an alarm occurred in the Water Treatment Plant 1, the operator would perform an operation on the monitoring and control device 30 to resolve the cause of the alarm, based on their past experience. In this case, the operator would perform an operation on the monitoring and control device 30 based on the operating conditions for that operation. An example of an operating condition is external factors such as weather. In the Water Treatment Plant 1, even for the same alarm, the operation to resolve the alarm may differ depending on external factors such as weather. Therefore, the operator would perform an operation on the monitoring and control device 30 to resolve the cause of the alarm, based on the type of alarm and external factors such as weather. That is, if the operating condition is weather, the operator would perform an operation on the monitoring and control device 30 to respond to the alarm, taking the weather into consideration. Details of the operating conditions will be described later.

[0023] If an alarm occurs in the water treatment plant 1 after the water treatment control support system 2 has been introduced, the operator will refer to the operation candidates 53 provided to the monitoring and control device 30 from the inference device 20 and execute an operation on the monitoring and control device 30 to resolve the cause of the alarm. The operation candidates 53 are candidates for operation based on the operating conditions 61B and the alarm information 63B when dealing with the alarm. Therefore, by executing the operation corresponding to the operation candidate 53 on the monitoring and control device 30, the operator will execute an operation on the monitoring and control device 30 that corresponds to the operating conditions 61B such as weather and the alarm information 63B.

[0024] When the data acquisition unit 11 acquires operation performance data 51A before the water treatment control support system 2 is introduced, it sends the operation performance data 51A to the learning model generation unit 12. Also, when the data acquisition unit 11 acquires operation performance data 51B after the water treatment control support system 2 is introduced, it sends the operation performance data 51B to the learning model generation unit 12.

[0025] Regardless of whether operation data 51A or 51B is acquired, the processing performed by the water treatment control support system 2 is the same. Below, we will describe the case in which the water treatment control support system 2 generates a trained model 50 based on operation data 51A.

[0026] The learning model generation unit 12 generates a trained model 50 for inferring the optimal operation candidate 53 corresponding to an alarm by performing training using multiple operation performance data 51A acquired by the data acquisition unit 11. The learning model generation unit 12 stores the generated trained model 50 in the storage device 60.

[0027] The inference device 20 comprises an inference information acquisition unit 21 and an inference unit 22. When an alarm occurs, the inference information acquisition unit 21 acquires the operation condition 61B corresponding to the generated alarm and the generated alarm information 63B, which is information about the generated alarm. The inference information acquisition unit 21 sends the acquired operation condition 61B and generated alarm information 63B to the inference unit 22. The operation condition 61B is used to infer the operation candidate 53, and the operation condition 61A is used to generate the trained model 50.

[0028] The inference unit 22 reads the trained model 50 from the storage device 60. The inference unit 22 also uses the trained model 50 generated by the trained model generation unit 12 to infer the optimal operation candidate 53 corresponding to the operating conditions 61B and the generated alarm information 63B when dealing with an alarm that is currently occurring. The inference unit 22 outputs the inferred operation candidate 53 to the monitoring control device 30. As a result, the monitoring control device 30 displays the operation candidate 53 on the monitoring screen of the display device 40. The operator inputs control commands corresponding to the water treatment control operations to the monitoring control device 30 while referring to the operation candidate 53. When an operation is performed by the operator, the monitoring control device 30 sends operation details 62B, which is information indicating the content of the operation, to the data acquisition unit 11.

[0029] Next, the learning phase by the learning device 10 will be explained using Figures 2 to 4, and the application phase by the inference device 20 will be explained using Figures 5 and 6.

[0030] <Learning Phase> Figure 2 is a diagram illustrating an example of the learning phase in a water treatment control support system according to an embodiment. In the learning phase, the learning device 10 generates a learned model 50. The data acquisition unit 11 of the learning device 10 acquires multiple operation performance data 51A.

[0031] Figure 2 shows a case where the learning device 10 generates a trained model 50 using operation performance data 51A, which shows past operation performance, as training data. That is, the learning device 10 generates a trained model 50 for inferring the optimal operation candidate 53 corresponding to the alarm that has occurred, based on the operation performance data 51A. Note that the learning device 10 uses multiple operation performance data 51A as training data, but only one operation performance data 51A is shown in Figure 2.

[0032] The learning device 10 learns the operation content performed by veteran operators as correct data. Specifically, the learning device 10 learns the operation content (correct data) 62B associated with inference data consisting of generated alarm information 63B and operation conditions 61B. The operation content (correct data) 62B performed by veteran operators may be weighted according to the veteran operator's years of experience. The operation performance data 51A consists of the first input, which is the operation conditions 61A and generated alarm information 63A, and the second input, which is the operation content 62A. That is, the combination of the operation conditions 61A, the generated alarm information 63A, and the operation content 62A is the learning data.

[0033] Operating condition 61A includes external factors when dealing with an alarm that has occurred, the operator's years of experience with the water treatment control support system 2, and the operating location where the water treatment control operation is performed by the operator.

[0034] The alarm indicated in the alarm information 63A is, for example, an alarm that occurred in the past. In this case, the external element is the external element corresponding to the alarm that occurred in the past, the years of experience is the years of experience of the operator who dealt with the alarm that occurred in the past at the time of dealing with the water treatment control support system 2, and the location of operation is the location where the operation was performed in response to the alarm that occurred in the past. The operation details 62A is information indicating the content of the operation performed by the operator on the monitoring and control device 30 when the alarm occurred.

[0035] The operating conditions 61A, the generated alarm information 63A, and the operation details 62A may be input to the data acquisition unit 11 at any time. The operating conditions 61A may be input to the data acquisition unit 11 after the alarm has occurred and the response to the alarm has been completed. Alternatively, the operating conditions 61A may be input to the data acquisition unit 11 after the alarm has occurred and the response to the alarm has been completed. Furthermore, the operation details 62A may be input to the data acquisition unit 11 after the operator has performed a water treatment control operation as a response to the alarm. Alternatively, the operation details 62A may be input to the data acquisition unit 11 in real time when the operator has performed a water treatment control operation as a response to the alarm.

[0036] Alarm information 63A indicates the type of alarm, i.e., the content of the alarm, for an abnormality or malfunction that occurred in water treatment plant 1. Examples of external factors include the season (spring, summer, autumn, winter, etc.), day of the week (weekday, Saturday, Sunday, etc.), time of day (morning, noon, evening, night, etc.), and weather (sunny, cloudy, rainy, heavy rain, etc.) when dealing with the alarm indicated in alarm information 63A. External factors such as season, day of the week, time of day, and weather have a significant impact on water usage, so the response to alarms, especially in water supply systems, may vary greatly depending on these external factors. External factors include at least one of the following: season, day of the week, time of day, and weather.

[0037] The years of experience refers to the length of service (e.g., junior, veteran) of the operator dealing with the alarm indicated in the alarm information 63A. The location of operation refers to the location (e.g., remote, central, on-site) of the monitoring and control device 30 when dealing with the alarm. External factors, years of experience, and location of operation are all pieces of information that influence the operation performed in response to the alarm. The operating conditions 61A include at least one of the external factors, years of experience, and location of operation, as well as the alarm information 63A.

[0038] Operation details 62A is information indicating the content of operations (operation history) performed by the operator in response to the conditions of operation 61A. Specifically, operation details 62A are the content of operations corresponding to external factors, years of experience, and location of operation when dealing with the alarm information 63A and the alarm indicated in the alarm information 63A. Operation details 62A shows the operation history actually performed by the operator (e.g., deployment screen, control operations on the deployment screen). Operation candidates 53 contain information of the same type as operation details 62A.

[0039] The data acquisition unit 11 acquires operating conditions 61A, occurrence alarm information 63A, and operation details 62A. Specifically, the data acquisition unit 11 acquires occurrence alarm information 63A from the water treatment equipment of the water treatment plant 1. The data acquisition unit 11 also acquires external elements from websites, computers installed in the water treatment plant 1, etc. For example, the data acquisition unit 11 acquires the weather at the location where the water treatment plant 1 is located from a website.

[0040] Furthermore, the data acquisition unit 11 acquires the operator's years of experience from the monitoring and control device 30. The monitoring and control device 30 stores the operator's username and years of experience in association with each other, and sends the years of experience corresponding to the username entered by the operator when they log in to the data acquisition unit 11.

[0041] Furthermore, the data acquisition unit 11 acquires the operation location, which is the location where the monitoring and control device 30 is located, using location information detected by GPS (Global Positioning System). The data acquisition unit 11 stores information indicating the location where each monitoring and control device 30 is located (operation location) and whether the monitoring and control device 30 at each location is a device that performs remote monitoring, central monitoring, or on-site monitoring. Based on the location information detected by GPS, the data acquisition unit 11 determines whether the monitoring and control device 30 that was operated is located remotely, centrally, or on-site. In this way, the data acquisition unit 11 acquires the operation location based on the location information detected by GPS. The data acquisition unit 11 sends the operation conditions 61A, the generated alarm information 63A, and the operation details 62A to the learning model generation unit 12.

[0042] The learning model generation unit 12 learns operation candidates 53 corresponding to operation performance data 51A. Specifically, the learning model generation unit 12 learns multiple operation performance data 51A and generates a trained model 50 for inferring the optimal operation candidate 53 corresponding to the operation conditions 61A and generated alarm information 63A when dealing with an alarm. In other words, the learning model generation unit 12 learns by using the operation content 62A, which is the second input, as the correct answer for the operation conditions 61A and generated alarm information 63A, which are the first inputs. Specifically, the learning model generation unit 12 learns by using the operation content 62A actually performed by the operator when dealing with an alarm that occurred in the past, as the correct answer for the generated alarm information 63A of an alarm that occurred in the past, and for the external factors, years of experience, and location of operation when dealing with this alarm.

[0043] The learning model generation unit 12 generates a trained model 50 that can output the optimal operation candidate 53 from the external factors, years of experience, and operation location given as the occurrence alarm information 63B and operation conditions 61B. Thus, the trained model 50 is a trained model that can output the operation candidate 53 corresponding to the operation conditions 61B and occurrence alarm information 63B when the operation conditions 61B and occurrence alarm information 63B are input. The operation conditions 61B and occurrence alarm information 63B are inference data for inferring the operation candidate 53.

[0044] The learning model generation unit 12 stores the generated trained model 50 in the storage device 60. The learning model generation unit 12 may also output the generated trained model 50 to the inference device 20. If the learning device 10 and the inference device 20 are implemented on the same computer, the learning model generation unit 12 may store the trained model 50 in a specific memory area.

[0045] The learning model generation unit 12 can use known algorithms such as supervised learning, unsupervised learning, and reinforcement learning as learning algorithms. As an example, we will explain the case where a neural network is applied to the learning algorithm used by the learning model generation unit 12.

[0046] The learning model generation unit 12 learns the optimal operation candidate 53 corresponding to the generated alarm information 63A and the operation condition 61A by so-called supervised learning, for example, according to a neural network model. Here, supervised learning is a method that learns the features contained in the learning data by providing the learning device 10 with pairs of data (training data) of input and result (labels), and infers the result from the input.

[0047] A neural network consists of an input layer made up of multiple neurons, a hidden layer (intermediate layer) made up of multiple neurons, and an output layer made up of multiple neurons. The hidden layer may be one layer or two or more layers.

[0048] Figure 3 is a diagram illustrating the neural network used in the learning device according to this embodiment. For example, in a three-layer neural network as shown in Figure 3, when multiple inputs are input to the input layer (X1~X3), the values ​​are multiplied by weights W1 (w11~w16) and input to the hidden layer (Y1~Y2). Then, the result is further multiplied by weights W2 (w21~w26) and output from the output layer (Z1~Z3). This output result varies depending on the values ​​of weights W1 and W2. In this embodiment, one such neural network is used.

[0049] The neural network used by the learning device 10 in Figure 2 learns the optimal operation candidate 53 corresponding to the operation conditions 61B and the generated alarm information 63B when dealing with an alarm, through so-called supervised learning, according to the operation performance data 51A created based on the combination of the first input and the second input (correct answer) acquired by the data acquisition unit 11.

[0050] In other words, a neural network learns by adjusting weights W1 and W2 so that when a first input is input to the input layer, the result output from the output layer approaches the second input (the correct answer).

[0051] The learning model generation unit 12 generates and outputs a trained model 50 by performing the learning described above. In this embodiment, an example in which the learning model generation unit 12 uses a supervised learning algorithm has been described, but the learning algorithm used by the learning model generation unit 12 is not limited to supervised learning algorithms. The learning model generation unit 12 can also apply reinforcement learning algorithms, unsupervised learning algorithms, semi-supervised learning algorithms, etc., in addition to supervised learning algorithms.

[0052] Furthermore, the learning model generation unit 12 may use operation performance data 51A selected by multiple monitoring and control devices 30 used in the same water treatment control support system 2 as learning data. Alternatively, the learning model generation unit 12 may use operation performance data 51A based on operation conditions 61A and generated alarm information 63A selected by monitoring and control devices 30 operating independently in different water treatment control support systems 2 as learning data.

[0053] Furthermore, during renovation work on the water treatment control support system 2, the monitoring and control device 30 for collecting learning data may be added to the water treatment control support system 2 midway through the process, or removed later. In addition, a learning device 10 that has learned the optimal operation candidates 53 corresponding to multiple operating conditions 61A and generated alarm information 63A when dealing with an alarm in a certain water treatment control support system 2 may be applied to another water treatment control support system 2, and the optimal operation candidates 53 corresponding to the operating conditions 61A and generated alarm information 63A when dealing with an alarm in that other water treatment control support system 2 may be relearned and updated.

[0054] Furthermore, the learning algorithm of the learning model generation unit 12 can also be deep learning, which learns to extract features themselves. In addition, the learning model generation unit 12 may perform machine learning according to other known methods, such as genetic programming, functional logic programming, or support vector machines.

[0055] Next, using Figure 4, the processing procedure for the learning device 10 to learn the trained model 50 will be described. Figure 4 is a flowchart showing the processing procedure for the learning process executed by the learning device according to the embodiment.

[0056] The data acquisition unit 11 acquires operation performance data 51A (step S1). Here, the data acquisition unit 11 simultaneously acquires operation conditions 61A, alarm information 63A, and operation details 62A as operation performance data 51A, but it is sufficient if the operation conditions 61A, alarm information 63A, and operation details 62A are input in association with each other. For this reason, the data acquisition unit 11 may acquire the operation conditions 61A, alarm information 63A, and operation details 62A at different times. The data acquisition unit 11 sends the operation conditions 61A, alarm information 63A, and operation details 62A to the learning model generation unit 12.

[0057] The learning model generation unit 12 performs a learning process using the operation conditions 61A, the alarm information 63A, and the operation content 62A (step S2). Specifically, the learning model generation unit 12 generates a trained model 50 by so-called supervised learning according to the training data created based on the combination of the operation conditions 61A, the alarm information 63A, and the operation content 62A. That is, the learning model generation unit 12 learns the relationship between the alarm information 63A of alarms that have occurred in the past, the external factors, years of experience, and location of operation when dealing with the alarm that occurred in the past, and the operation history (operation content 62A) corresponding to the alarm that occurred in the past. Through this learning process, the learning model generation unit 12 generates a trained model 50 for inferring the optimal operation candidate 53 corresponding to the operation conditions 61A and the alarm information 63A when dealing with the alarm that occurred.

[0058] After generating the trained model 50, the learning model generation unit 12 outputs the trained model 50 to the storage device 60 or the inference device 20 (step S3).

[0059] <Utilization Phase> Figure 5 is a diagram illustrating an example of the utilization phase in the water treatment control support system according to the embodiment. In the utilization phase, the inference device 20 acquires the operation conditions 61B and the occurrence alarm information 63B, which are inference data. The inference device 20 then calculates the operation candidate 53 using the trained model 50 and the operation conditions 61B and occurrence alarm information 63B, and outputs the operation candidate 53 to the monitoring and control device 30. The operation conditions 61B are the operation conditions when an alarm occurs, and, like the operation conditions 61A, include external factors, years of experience, and the location of operation. The inference device 20 acquires the operation conditions 61B and occurrence alarm information 63B in the same manner as the learning device 10 acquires the operation conditions 61A and occurrence alarm information 63A.

[0060] When an alarm occurs, the inference information acquisition unit 21 acquires the corresponding operating conditions 61B and the generated alarm information 63B. The operating conditions 61B include, as a first input, external factors, years of experience, and location of operation when dealing with the generated alarm.

[0061] The inference unit 22 uses the trained model 50 to infer the optimal operation candidate 53 corresponding to the external factors, years of experience, and location of operation when dealing with the alarm that occurred, as indicated by the alarm information 63B and the operation conditions 61B. The inference unit 22 outputs the optimal operation candidate 53 corresponding to the alarm information 63B to the monitoring and control device 30 as a first output.

[0062] The inference unit 22 calculates the operation candidates 53, which include the display screen to be displayed on the monitoring and control device 30, and the control operations to be performed by the operator on the display screen. The inference unit 22 sends the calculated operation candidates 53 to the monitoring and control device 30.

[0063] The monitoring and control device 30 displays the operation candidates 53 sent from the inference unit 22 on the screen of the display device 40. Specifically, the monitoring and control device 30 displays the expanded screen included in the operation candidates 53 on the display device 40. Furthermore, the monitoring and control device 30 displays the control operations on the expanded screen included in the operation candidates 53 on the display device 40.

[0064] The display device 40 displays the deployment screen and the control operations that the operator should perform on the deployment screen as candidate operations 53 for dealing with the alarm that has occurred. The operator performs the operation corresponding to the alarm that has occurred on the monitoring and control device 30, referring to the control operations displayed on the display device 40 on the deployment screen displayed on the display device 40.

[0065] Next, using Figure 6, the processing procedure for the inference device 20 to infer the operation candidate 53 will be explained. Figure 6 is a flowchart showing the processing procedure for the inference process executed by the inference device according to the embodiment.

[0066] When an alarm occurs, the inference information acquisition unit 21 acquires the corresponding operating conditions 61B and the generated alarm information 63B (step S11). The inference information acquisition unit 21 sends the operating conditions 61B and the generated alarm information 63B to the inference unit 22.

[0067] The inference unit 22 receives the operation conditions 61B and the occurrence alarm information 63B sent from the inference information acquisition unit 21. The inference unit 22 also reads the trained model 50 from the storage device 60. The inference unit 22 infers an operation candidate 53 using the trained model 50 (step S12). That is, the inference unit 22 applies the operation conditions 61B and the occurrence alarm information 63B to the trained model 50 to obtain the optimal operation candidate 53 (first output). The inference unit 22 outputs the optimal operation candidate 53 to the monitoring and control device 30 (step S13).

[0068] The monitoring and control device 30 displays the deployment screen and control operation included in the operation candidate 53 on the display device 40. The operator inputs a control command corresponding to the displayed control operation to the monitoring and control device 30 while referring to the displayed control operation. The monitoring and control device 30 generates operation content 62B that shows the control operation corresponding to the input control command and the deployment screen on which the control command was input, and sends it to the data acquisition unit 11. The data acquisition unit 11 acquires the operation conditions 61B and the generated alarm information 63B corresponding to the operation content 62B. The combination of the operation content 62B and the operation conditions 61B and generated alarm information 63B corresponding to the operation content 62B is the operation performance data 51B. The learning device 10 performs further learning using the operation performance data 51B.

[0069] In the water treatment control support system 2, the learning device 10 repeatedly performs the process of learning a trained model 50 using operation performance data 51B, the inference device 20 calculates operation candidates 53 using the trained model 50 and presents them to the operator, and the learning device 10 repeatedly performs the process of acquiring operation performance data 51B corresponding to the operations performed by the operator.

[0070] As described above, according to this embodiment, the water treatment control support system 2 can efficiently generate optimal operation candidates 53 corresponding to the operating conditions 61B and the generated alarm information 63B when dealing with an alarm that has occurred.

[0071] Specifically, the water treatment control support system 2 of this embodiment can expedite operations by showing the operator the optimal operation candidate 53 corresponding to the operating conditions 61B and the generated alarm information 63B when dealing with an alarm that has occurred. As a result, the water treatment control support system 2 can improve the efficiency and speed of plant operation, and can also improve the accuracy of operations when dealing with an alarm that has occurred, thereby realizing safe plant operation.

[0072] Furthermore, the water treatment control support system 2 of this embodiment includes the operator's years of experience as input data to the learned model 50. Therefore, it can output operation candidates 53 to the monitoring and control device 30 that take into account the differences in operations when dealing with alarms that arise due to differences in years of experience. Consequently, the water treatment control support system 2 can present the optimal operation candidate 53 for each operator.

[0073] Young operators with limited experience often respond to alarms by displaying a graphic screen that simulates the plant's status, and then operating equipment (such as starting or stopping pumps) based on an overview of the situation on the graphic screen.

[0074] On the other hand, experienced veteran operators often display a graphic screen to grasp the overall situation, and then display a parameter adjustment screen to change the values ​​of various parameters used in PID (Proportional-Integral-Differential) control, etc., in order to perform detailed setting changes in order to perform the optimal operation according to the situation. Thus, even when the same alarm occurs, the operation may differ depending on the operator's years of experience. For this reason, the water treatment control support system 2 of this embodiment includes the operator's years of experience as input data to the learned model 50, making it possible to present optimal operation candidates 53 for each operator, taking into account their years of experience.

[0075] Furthermore, since the water treatment control support system 2 of this embodiment includes the operating location of the monitoring and control device 30 as input data to the learned model 50, it can output operation candidates 53 to the monitoring and control device 30 that take into account the differences in operation when dealing with alarms caused by differences in operating location. Therefore, the water treatment control support system 2 can present the optimal operation candidate 53 for each operating location.

[0076] For example, in water and wastewater treatment plants, there are mainly three types of monitoring methods: remote monitoring, central monitoring, and on-site monitoring. Depending on the difference in these monitoring methods, the purpose of monitoring and control may also differ. For example, remote monitoring aims to monitor the overall plant operation status without the aim of fine-grained control, so it generally does not display detailed control screens, and instead displays an overall monitoring screen. On the other hand, in central monitoring and on-site monitoring, detailed control operations are often performed on control screens according to the situation in order to ensure safe plant operation. Thus, the content of operations when dealing with an alarm that occurs may differ depending on the location of operation in the plant. For this reason, the water treatment control support system 2 of this embodiment includes the location of operation of the monitoring and control device 30 as input data for the learned model 50, making it possible to present the optimal operation candidate 53 for each operation location that takes the location into consideration.

[0077] The water treatment control support system 2 generates a learned model 50 using the operation conditions 61A for alarms and the operation content 62A for each occurrence of alarm information 63A. By inputting the operation conditions 61B and the occurrence of alarm information 63B into the learned model 50, it calculates operation candidates 53 corresponding to the operation conditions 61B and the occurrence of alarm information 63B. As a result, the water treatment plant 1 can provide operators with operation candidates 53 corresponding to the operation conditions 61B and the occurrence of alarm information 63B, taking into account various complex operating conditions such as external factors, years of experience, and operating location, without having to pre-register the operation content for each occurrence of the operation conditions 61B and the occurrence of alarm information 63B. Furthermore, since the water treatment control support system 2 presents not only a monitoring screen (expanded screen) corresponding to the alarm but also control operation candidates, operators can easily perform operations corresponding to alarms while referring to the control operation candidates.

[0078] Next, the hardware configuration of the water treatment control support system 2 will be described. In the water treatment control support system 2, the learning device 10, the memory device 60, and the inference device 20 are implemented by processing circuits. These processing circuits may be a processor and memory that execute programs stored in memory, or they may be dedicated hardware. Processing circuits are also called control circuits.

[0079] Figure 7 shows an example of the configuration of a processing circuit when the processing circuit of the water treatment control support system according to the embodiment is realized with a processor and memory. The processing circuit 90 shown in Figure 7 is a control circuit and comprises a processor 91 and memory 92. When the processing circuit 90 is composed of a processor 91 and memory 92, each function of the processing circuit 90 is realized by software, firmware, or a combination of software and firmware. The software or firmware is written as a program and stored in memory 92. In the processing circuit 90, each function is realized by the processor 91 reading and executing the program stored in memory 92. In other words, the processing circuit 90 includes memory 92 for storing a water treatment control support program, which will result in the processing of the water treatment control support system 2 being executed. This water treatment control support program can also be said to be a program that causes the water treatment control support system 2 to execute each function realized by the processing circuit 90. This water treatment control support program may be provided by a storage medium on which the program is stored, or by other means such as a communication medium.

[0080] Here, the processor 91 is, for example, a CPU (Central Processing Unit), processing unit, arithmetic unit, microprocessor, microcomputer, or DSP (Digital Signal Processor). The memory 92 is, for example, a non-volatile or volatile semiconductor memory such as RAM (Random Access Memory), ROM (Read Only Memory), flash memory, EPROM (Erasable Programmable ROM), EEPROM (Registered Trademark) (Electrically EPROM), magnetic disks, flexible disks, optical disks, compact disks, minidiscs, or DVDs (Digital Versatile Discs).

[0081] Figure 8 shows an example of a processing circuit when the processing circuit of the water treatment control support system according to the embodiment is configured with dedicated hardware. The processing circuit 93 shown in Figure 8 may be, for example, a single circuit, a composite circuit, a programmed processor, a parallel programmed processor, an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array), or a combination thereof.

[0082] The processing circuits 90 and 93 may be partially implemented with dedicated hardware and partially with software or firmware. In this way, the processing circuits 90 and 93 can realize the above-mentioned functions with dedicated hardware, software, firmware, or a combination thereof. The learning device 10 and the inference device 20 may be implemented with separate processing circuits.

[0083] As described above, the water treatment control support system 2 includes operating conditions 61A such as external factors, years of experience, and operating location, as well as generated alarm information 63A, as input data to the learned model 50. This makes it possible to learn operation content 62A corresponding to the operating conditions 61A and generated alarm information 63A. As a result, the water treatment control support system 2 can calculate operation candidates 53 according to the operating conditions 61B and generated alarm information 63B, and thus can provide operation support to operators according to the operating conditions 61B and generated alarm information 63B.

[0084] The configurations shown in the above embodiments are merely examples, and can be combined with other known technologies. It is also possible to omit or modify parts of the configuration without departing from the gist of the invention. [Explanation of Symbols]

[0085] 1 Water treatment plant, 2 Water treatment control support system, 10 Learning device, 11 Data acquisition unit, 12 Learning model generation unit, 20 Inference device, 21 Inference information acquisition unit, 22 Inference unit, 30 Monitoring and control device, 40 Display device, 50 Learned model, 51A, 51B Operation performance data, 53 Operation candidate, 60 Storage device, 61A, 61B Operation conditions, 62A, 62B Operation details, 63A, 63B Occurred alarm information, 90, 93 Processing circuit, 91 Processor, 92 Memory.

Claims

1. The water treatment plant has an inference device that infers candidate operations, which are candidates for operations when an alarm occurs. The inference device is An inference information acquisition unit acquires, as inference data, alarm occurrence information indicating the type of alarm and operating conditions indicating conditions that affect the operation to deal with the alarm. An inference unit inputs the inference data into a trained model for inferring operation candidates from the inference data, infers operation candidates corresponding to the occurrence alarm information and the operation conditions from the occurrence alarm information and the operation conditions, and outputs the inferred operation candidates. Equipped with, The aforementioned operating conditions include external factors when dealing with the alarm, The aforementioned external factors include at least one of the season, day of the week, time of day, and weather conditions at the time of responding to the alarm. A water treatment control support system characterized by the following features.

2. The water treatment plant has a learning device that learns the operations to be performed when the alarm occurs, The learning device is A data acquisition unit that acquires the aforementioned alarm information, the aforementioned operating conditions, and the operation details indicating the content of the operation when the alarm occurred as learning data. A learning model generation unit that generates the trained model using the aforementioned training data, Equipped with, The water treatment control support system according to feature 1.

3. A monitoring and control device that monitors the water treatment plant and controls the water treatment plant in accordance with the operations performed by the operator, A display device that displays information sent from the aforementioned monitoring and control device, Furthermore, The inference unit outputs the operation candidate to the monitoring and control device, The monitoring and control device causes the operation candidates to be displayed on the display device. The water treatment control support system according to claim 1 or 2.

4. The aforementioned operating conditions include the number of years of experience of the operator of the monitoring and control device who takes action against the alarm. The water treatment control support system according to feature 3.

5. The aforementioned operating conditions include the operating location in which the monitoring and control device is operated. The water treatment control support system according to feature 3 or 4.

6. The operation candidates include a display screen to be shown on the display device and control operations on the display screen. The water treatment control support system according to feature 3.

7. The system further includes a monitoring and control device that monitors the water treatment plant and controls the water treatment plant in accordance with the operations performed by the operator, When the monitoring and control device is operated in the event of the alarm, it controls the water treatment plant and inputs the operation details, which indicate the operation performed, into the data acquisition unit. The water treatment control support system according to feature 2.

8. The inference step includes inferring candidate operations, which are candidates for operations to be performed when an alarm occurs in a water treatment plant. The aforementioned inference step is, An inference information acquisition step involves acquiring alarm occurrence information indicating the type of alarm and operation conditions indicating conditions that affect the operation to deal with the alarm, as inference data. An inference step in which the inference data is input to a trained model for inferring operation candidates from the inference data, thereby inferring operation candidates corresponding to the occurrence alarm information and the operation conditions from the occurrence alarm information and the operation conditions, and outputting the inferred operation candidates; Includes, The aforementioned operating conditions include external factors when dealing with the alarm, The aforementioned external factors include at least one of the season, day of the week, time of day, and weather conditions at the time of responding to the alarm. A water treatment control support method characterized by the following:

9. The water treatment plant includes a learning step to learn the operation to be performed when the alarm occurs, The aforementioned learning steps are: A data acquisition step of acquiring the aforementioned alarm information, the operating conditions, and the operation details indicating the content of the operation when the alarm occurred as training data, A learning model generation step in which the trained model is generated using the aforementioned training data, including, The water treatment control support method according to feature 8.

10. The computer is instructed to perform an inference step to infer candidate actions that can be taken when an alarm occurs in a water treatment plant. An inference information acquisition step involves acquiring alarm occurrence information indicating the type of alarm and operation conditions indicating conditions that affect the operation to deal with the alarm, as inference data. An inference step in which the inference data is input to a trained model for inferring operation candidates from the inference data, thereby inferring operation candidates corresponding to the occurrence alarm information and the operation conditions from the occurrence alarm information and the operation conditions, and outputting the inferred operation candidates; Includes, The aforementioned operating conditions include external factors when dealing with the alarm, The aforementioned external factors include at least one of the season, day of the week, time of day, and weather conditions at the time of responding to the alarm. A water treatment control support program characterized by the following features.

11. The computer is instructed to perform a learning step to learn the operation to be performed when the alarm occurs in the water treatment plant. The aforementioned learning steps are: A data acquisition step of acquiring the aforementioned alarm information, the operating conditions, and the operation details indicating the content of the operation when the alarm occurred as training data, A learning model generation step in which the trained model is generated using the aforementioned training data, including, The water treatment control support program according to feature 10.

Images