Operation value output device, operation value output method, and program

Abstract

This technology enables the operation of a sewage treatment plant in a manner that does not exceed the standards for effluent quality set for sewage treatment facilities. [Solution] The operation value output device (1) includes: an effluent water quality estimation unit (133) that estimates future effluent water quality data based on at least past water quality data; a setting unit (14) that sets target values ​​for water quality data and operating costs; and an operation unit (15) that inputs the target values ​​set by the setting unit (14), the water quality data acquired by the acquisition unit (10), and the future effluent water quality data estimated by the effluent water quality estimation unit (133) into a learned operation model and outputs operation values ​​for the sewage treatment plant (2).

Description

Technical Field

[0001] The present disclosure relates to an operation operation value output device, an operation operation value output method, and a program.

Background Art

[0002] According to Patent Document 1, a system that can operate and control a water treatment plant based on accumulated data from the past to the present (water quality data, weather information, seasonal information), and future influent water quality (including water volume) estimated from three types of data: weather information, seasonal information, and past influent water quality data from the past to the present and into the future is disclosed.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] However, in the system disclosed in Patent Document 1, since the water treatment plant is controlled based on the influent water quality, there is a possibility of exceeding the standard of the treated water quality defined for each water treatment facility.

Means for Solving the Problems

[0005] To solve the above problems, an operating value output device according to one aspect of the present disclosure includes: an acquisition unit for acquiring water quality data in a sewage treatment plant; a plant model which is a simulator that models the processes of the sewage treatment plant and analyzes water quality data from the operating values ​​of the sewage treatment plant; an operating model learning unit which reinforces an operating model that outputs the operating values ​​based on the water quality data and operating costs analyzed by the plant model; an effluent water quality estimation unit which estimates future effluent water quality data based on at least past water quality data; a setting unit for setting target values ​​for water quality data and operating costs; and an operating unit which inputs the target values ​​set by the setting unit, the water quality data acquired by the acquisition unit, and the future effluent water quality data estimated by the effluent water quality estimation unit into the learned operating model and outputs the operating values ​​of the sewage treatment plant.

[0006] To solve the above problems, a method for outputting operational values ​​according to one aspect of the present disclosure includes the steps of: acquiring water quality data in a sewage treatment plant; reinforcing a power operation model that outputs operational values ​​based on water quality data and operating costs analyzed by a plant model, which is a simulator that models the processes of the sewage treatment plant and analyzes water quality data from operational values ​​of the sewage treatment plant; estimating future effluent water quality data based on at least past water quality data; setting target values ​​for water quality data and operating costs; and inputting the set target values, the acquired water quality data, and the estimated future effluent water quality data into the trained power operation model to output operational values ​​of the sewage treatment plant.

[0007] To solve the above problems, a program according to one aspect of the present disclosure causes a computer to perform the following: a process of acquiring water quality data in a sewage treatment plant; a process of reinforcing an operation model, which is a simulator that models the processes of the sewage treatment plant and outputs operation values ​​based on water quality data analyzed by a plant model that analyzes water quality data from the operation values ​​of the sewage treatment plant; a process of estimating future effluent water quality data based on at least past water quality data; a process of setting target values ​​for water quality data and operation costs; and a process of inputting the set target values, the acquired water quality data, and the estimated future effluent water quality data into the trained operation model and outputting the operation values ​​of the sewage treatment plant. [Effects of the Invention]

[0008] According to one aspect of this disclosure, the operation of a sewage treatment plant can be carried out in such a way as not to exceed the standards for effluent quality set for sewage treatment facilities. [Brief explanation of the drawing]

[0009] [Figure 1] This figure shows an example configuration of a sewage treatment system including an operating value output device according to Embodiment 1 of this disclosure. [Figure 2] This is a flowchart illustrating the processing procedure of the operation value output device according to Embodiment 1 of this disclosure. [Figure 3] This figure shows an example configuration of a sewage treatment system including an operating value output device according to Embodiment 2 of this disclosure. [Figure 4] This is a flowchart illustrating the processing procedure of the operation value output device according to Embodiment 2 of this disclosure. [Modes for carrying out the invention]

[0010] Hereinafter, embodiments relating to one aspect of this disclosure (hereinafter also referred to as "this embodiment") will be described based on the drawings. In the drawings, the same or corresponding parts are denoted by the same reference numerals, and their descriptions will not be repeated.

[0011] [Embodiment 1] (Configuration of sewage treatment system 100) Figure 1 is a block diagram showing the configuration of the main parts of the sewage treatment system 100 according to this embodiment. The sewage treatment system 100 includes an operation value output device 1 and a sewage treatment plant 2.

[0012] Sewage treatment plant 2 is equipped with a primary sedimentation tank, reaction tank, final sedimentation tank, blower, airflow control valve, aeration device, pump, measuring instruments, gravity thickening tank, mechanical thickening tank, digester, dewatering tank, piping, etc. Since sewage treatment plant 2 itself has a typical configuration, a detailed explanation will not be provided here.

[0013] The measuring instruments in wastewater treatment plant 2 are installed to measure the water quality on the inflow side and the discharge side. The water quality data measured by the instruments include suspended solids (SS), activated sludge sedimentation rate (SV), activated sludge suspended solids (MLSS), total phosphorus, total nitrogen, biochemical oxygen demand (BOD), chemical oxygen demand (COD), pH value, and dissolved oxygen (DO).

[0014] As shown in Figure 1, the operation value output device 1 includes an acquisition unit 10, a water quality data storage unit 11, an operation model generation unit 12, a water quality estimation unit 13, a setting unit 14, and an operation unit 15.

[0015] The acquisition unit 10 periodically acquires inflow water quality data and effluent water quality data from measuring instruments installed in the sewage treatment plant 2 and stores them in the water quality data storage unit 11. The inflow water quality data and effluent water quality data are collectively referred to as water quality data.

[0016] The water quality data storage unit 11 is a database that stores various water quality data regularly acquired by the acquisition unit 10. In addition to various water quality data, weather information and season information are also stored.

[0017] As shown in FIG. 1, the operation control model generation unit 12 includes a plant model 121, an operation control model learning unit 122, and an operation control model storage unit 123. The plant model 121 is a simulator that models the process of the sewage treatment plant 2, and is constructed by the physical characteristics of the sewage treatment plant 2 and calibration based on the water quality data accumulated in the water quality data storage unit 11.

[0018] The operation control model learning unit 122 learns the optimal operation control by learning the operation control model through reinforcement learning using the plant model 121. The operation control value analyzed by the operation control model of the operation control model learning unit 122 is input to the plant model 121. Then, the water quality data obtained by the simulation in the plant model 121 is input to the operation control model. The operation control model obtains an optimal operation control by obtaining a predefined reward according to the water quality data input to the operation control model and the operation cost by the operation control model.

[0019] The operation control model is constructed by a reinforcement learning model. In the Markov decision process, the environment at time t takes the state st, and the agent selects an available action at in the state st. Then, the environment transitions to a new state st+1, and at this time the agent receives a reward rt or rt+1 corresponding to the state transition.

[0020] The water quality of the sewage treatment plant 2 in the state may be the discharged water quality at the current time, or may include the predicted value of the past or future discharged water quality in the discharged water quality at the current time, and may include weather information or season information as the state of the target system as needed.

[0021] The action guidelines and action rules, which are strategies, are determined by the Q-table in the case of Q-learning, which is a typical method of reinforcement learning, and are determined by a neural network in the case of deep reinforcement learning. Actions include the inflow rate, which is the amount of water from the initial sedimentation tank to the reaction tank, the blowing rate, which is the amount of air supplied into the reaction tank or the DO target value, the return sludge amount, which is the amount of sludge returned to the reaction tank, the excess sludge withdrawal amount, which is the amount of sludge withdrawn from the final sedimentation tank, the flocculant injection rate, and the like. Rewards are appropriately set according to water quality (total nitrogen, total phosphorus, SV, BOD, etc.) and operating costs (the reciprocal of the power consumption generated during operation).

[0022] The operation operation model storage unit 123 stores the parameters of the operation operation model learned by the operation operation model learning unit 122 (for example, in the case of a neural network, parameters such as the weights between each node).

[0023] As shown in FIG. 1, the water quality estimation unit 13 includes a treated water quality estimation model learning unit 131, a treated water quality estimation model storage unit 132, and a treated water quality estimation unit 133.

[0024] Based on the weather information from the past to the present and into the future, the seasonal information, and the past treated water quality data among the various water quality data stored in the water quality data storage unit 11, the treated water quality estimation model learning unit 131 constructs and learns a treated water quality estimation model for estimating the water quality that the sewage treatment facility will discharge in the future. Here, the learning is performed using the past operation data stored in the water quality data storage unit 11.

[0025] The treated water quality estimation model includes a feature extraction layer, a LSTM (Long Short-Term Memory) layer, and an output layer. The weather information and the treated water quality data at that time are input into the feature extraction layer. Here, the weather information includes the weather information and the seasonal information, and examples of the weather information include mesh weather data with latitude on the vertical axis and longitude on the horizontal axis.

[0026] Furthermore, if the relevant time is in the future, after time t, weather forecast data is used as weather information. The feature extraction layer extracts the features of this mesh weather data using a convolutional layer and inputs them together with the discharge water quality data into a fully connected layer to output the final features. The output of the feature extraction layer is input to the LSTM layer. The LSTM layer holds the information from the encoder section and reflects it in the output of the decoder section.

[0027] The output layer of the decoder unit performs processing such as a fully connected layer to estimate the effluent water quality for the next time step. The estimated effluent water quality is input to the feature extraction layer for the next time step. The decoder unit is set up in a sequence corresponding to the number of time steps to be predicted. For example, if one time step is one hour and it is desired to predict the effluent water quality up to 24 hours in advance, 25 decoder units will be set up in a sequence, covering the period from the current time t to time t+24 (24 hours later).

[0028] The discharge water quality estimation model storage unit 132 stores the trained parameters (parameters such as weights between each node) of the discharge water quality estimation model obtained through training by the discharge water quality estimation model learning unit 131.

[0029] The discharge water quality estimation unit 133 uses a discharge water quality estimation model constructed from parameters stored in the discharge water quality estimation model storage unit 132 to estimate the water quality to be discharged in the future based on various water quality data stored in the water quality data storage unit 11, and outputs the estimated discharge water quality value to the operation control unit 15.

[0030] As shown in Figure 1, the setting unit 14 includes a setting input unit 141 and a setting storage unit 142. The setting input unit 141 receives setting values ​​from the operator. Here, the setting values ​​are target values ​​for effluent water quality and operating costs set by the operator. Examples of effluent water quality include activated sludge sedimentation rate (SV) and biochemical oxygen demand (BOD). The setting storage unit 142 stores the setting values ​​entered into the setting input unit 141 by the operator.

[0031] The operation control unit 15 reads the trained operation control model stored in the operation control model storage unit 123. The operation control unit 15 then operates the operation control model using the past and present water quality data input from the water quality data storage unit 11, the setting values ​​stored in the setting storage unit 142, and the discharge water quality estimate values ​​output by the water quality estimation unit 13 as input. The operation control unit 15 then outputs the optimal operation control values ​​to determine the operation control values ​​for controlling the sewage treatment plant 2.

[0032] (Processing flow of the operation value output device 1) Figure 2 is a flowchart illustrating the processing procedure of the operation value output device 1 according to Embodiment 1 of this disclosure. First, the operator inputs a setting value into the setting input unit 141 (S11). The setting value input into the setting input unit 141 by the operator is stored in the setting storage unit 142.

[0033] Next, the operation model learning unit 122 uses the plant model 121 to train the operation model through reinforcement learning, thereby teaching the operation model the optimal operation (S12).

[0034] Next, the effluent water quality estimation model learning unit 131 constructs and learns an effluent water quality estimation model that estimates the water quality to be discharged into the sewage treatment facility in the future, based on the various water quality data stored in the water quality data storage unit 11 (S13). Then, the effluent water quality estimation model learning unit 131 stores the learned parameters of the effluent water quality estimation model (parameters such as weights between each node) in the effluent water quality estimation model storage unit 132.

[0035] Next, the discharge water quality estimation unit 133 uses the discharge water quality estimation model constructed from the parameters stored in the discharge water quality estimation model storage unit 132 to estimate the water quality to be discharged in the future based on the various water quality data stored in the water quality data storage unit 11 (S14).

[0036] Next, the operation unit 15 reads the trained operation model stored in the operation model storage unit 123. Then, the operation unit 15 operates the operation model using the past and present water quality data input from the water quality data storage unit 11, the setting values ​​stored in the setting storage unit 142, and the effluent water quality estimate values ​​output by the water quality estimation unit 13 as input. Finally, the operation unit 15 determines the optimal operation values ​​for controlling the sewage treatment plant 2 (S15).

[0037] Next, the operation unit 15 controls the sewage treatment plant 2 by setting the determined operation values ​​to each piece of equipment in the sewage treatment plant 2 (S16). Finally, the operation value output device 1 determines whether or not to terminate the treatment (S17). If the treatment is not to be terminated (S17, No), the operation value output device 1 returns to step S12 and repeats the subsequent processing. If the treatment is to be terminated (S17, Yes), the operation value output device 1 terminates the treatment.

[0038] (Effect of the operation value output device 1) As described above, according to the operation value output device 1 of this embodiment, the operation unit 15 operates the operation model based on the effluent water quality estimate values ​​output by the water quality estimation unit 13. The operation unit 15 then determines the optimal operation values ​​for controlling the sewage treatment plant 2. Therefore, operation can be performed without exceeding the effluent water quality standards set for the sewage treatment facility.

[0039] Furthermore, monitoring effluent quality, compared to monitoring inflow water quality which has no standards, allows for control while monitoring effluent quality which does have standards, making it easier to control the system to stay within the standard values. As a result, it may be possible to reduce operating costs.

[0040] [Embodiment 2] Figure 3 is a block diagram showing the configuration of the main parts of the sewage treatment system 100A according to Embodiment 2 of this disclosure. The sewage treatment system 100A comprises an operation value output device 1A and a sewage treatment plant 2. Parts having the same configuration and function as the sewage treatment system 100 according to Embodiment 1 shown in Figure 1 are given the same reference numerals, and their detailed descriptions will not be repeated.

[0041] As shown in Figure 3, the operation value output device 1A according to Embodiment 2 of the present disclosure has an inflow water quality estimation model learning unit 134, an inflow water quality estimation model storage unit 135, and an inflow water quality estimation unit 136 added to the water quality estimation unit 13 of the operation value output device 1 according to Embodiment 1 of the present disclosure shown in Figure 1.

[0042] The inflow water quality estimation model learning unit 134 constructs and learns a learning model that estimates the amount and quality of water that will flow into the sewage treatment facility in the future, based on weather information from the past through to the present, seasonal information, and past inflow water quality data, among the various water quality data stored in the water quality data storage unit 11. Here, learning is performed using past operating data stored in the water quality data storage unit 101.

[0043] The inflow water quality estimation model storage unit 135 stores the learned parameters (parameters such as weights between each node) of the inflow water quality estimation model obtained through learning by the inflow water quality estimation model learning unit 134.

[0044] The inflow water quality estimation unit 136 uses an inflow water quality estimation model constructed from parameters stored in the inflow water quality estimation model storage unit 135 to estimate the future inflow volume and water quality based on various water quality data stored in the water quality data storage unit 11, and outputs the estimated inflow water quality value to the operation control unit 15A.

[0045] The operation control unit 15A reads the trained operation control model stored in the operation control model storage unit 123. The operation control unit 15A then operates the operation control model using the past and present water quality data input from the water quality data storage unit 11, the setting values ​​stored in the setting storage unit 142, and the inflow water quality estimates and effluent water quality estimates output by the water quality estimation unit 13 as input. The operation control unit 15A then determines the optimal operation control values ​​for controlling the sewage treatment plant 2.

[0046] (Processing flow of the operation value output device 1A) Figure 4 is a flowchart illustrating the processing procedure of the operation value output device 1A according to Embodiment 2 of this disclosure. First, the operator inputs a setting value into the setting input unit 141 (S21). The setting value input into the setting input unit 141 by the operator is stored in the setting storage unit 142.

[0047] Next, the operation model learning unit 122 uses the plant model 121 to train the operation model through reinforcement learning, thereby teaching the operation model the optimal operation (S22).

[0048] Next, the inflow water quality estimation model learning unit 134 constructs and learns an inflow water quality estimation model that estimates the quality of water that will flow into the sewage treatment facility in the future, based on the various water quality data stored in the water quality data storage unit 11 (S23). Then, the inflow water quality estimation model learning unit 134 stores the learned parameters of the inflow water quality estimation model (parameters such as weights between each node) in the inflow water quality estimation model storage unit 135.

[0049] Next, the effluent water quality estimation model learning unit 131 constructs and learns an effluent water quality estimation model that estimates the water quality that the sewage treatment facility will discharge in the future, based on the various water quality data stored in the water quality data storage unit 11 (S24). Then, the effluent water quality estimation model learning unit 131 stores the learned parameters of the effluent water quality estimation model (parameters such as weights between each node) in the effluent water quality estimation model storage unit 132.

[0050] Next, the inflow water quality estimation unit 136 uses the inflow water quality estimation model constructed from the parameters stored in the inflow water quality estimation model storage unit 135 to estimate the future water quality based on the various water quality data stored in the water quality data storage unit 11 (S25).

[0051] Next, the discharge water quality estimation unit 133 uses a discharge water quality estimation model constructed from parameters stored in the discharge water quality estimation model storage unit 132 to estimate the water quality to be discharged in the future based on various water quality data stored in the water quality data storage unit 11 (S26).

[0052] Next, the operation control unit 15A reads the trained operation control model stored in the operation control model storage unit 123. Then, the operation control unit 15A operates the operation control model using the past and present water quality data input from the water quality data storage unit 11, the setting values ​​stored in the setting storage unit 142, and the inflow water quality estimates and effluent water quality estimates output by the water quality estimation unit 13 as input. Finally, the operation control unit 15A determines the optimal operation control values ​​for controlling the sewage treatment plant 2 (S27).

[0053] Next, the operation unit 15A controls the sewage treatment plant 2 by setting the determined operation values ​​to each piece of equipment in the sewage treatment plant 2 (S28). Finally, the operation value output device 1A determines whether or not to terminate the treatment (S29). If the treatment is not to be terminated (S29, No), the operation value output device 1A returns to step S22 and repeats the subsequent processing. If the treatment is to be terminated (S29, Yes), the operation value output device 1A terminates the treatment.

[0054] (Effect of the operation value output device 1A) As described above, according to the operation value output device 1A of this embodiment, the operation unit 15A operates the operation model based on the inflow water quality estimate and effluent water quality estimate output by the water quality estimation unit 13. The operation unit 15 then determines the optimal operation values ​​for controlling the sewage treatment plant 2. Therefore, operation can be performed without exceeding the effluent water quality standards set for the sewage treatment facility.

[0055] [Examples of implementation using software] The control block of the operation value output device 1,1A may be implemented by logic circuits (hardware) formed on an integrated circuit (IC chip) or the like, or by software.

[0056] In the latter case, the operation value output device 1,1A includes a computer that executes instructions for a program, which is software that realizes each function. This computer includes, for example, one or more processors and a computer-readable recording medium that stores the program. The object of this disclosure is achieved when the processor reads the program from the recording medium and executes it in the computer. For example, a CPU (Central Processing Unit) can be used as the processor. As the recording medium, a "tangible medium that is not temporary," such as ROM (Read Only Memory), can be used, as well as tape, disk, card, semiconductor memory, programmable logic circuit, etc. It may also further include RAM (Random Access Memory) for deploying the program. Furthermore, the program may be supplied to the computer via any transmission medium capable of transmitting the program (such as a communication network or broadcast wave). One aspect of this disclosure can also be realized in the form of a data signal embedded in a carrier wave, in which the program is embodied by electronic transmission.

[0057] 〔summary〕 The operation value output device according to Embodiment 1 of the present disclosure comprises: an acquisition unit for acquiring water quality data in a sewage treatment plant; a plant model which is a simulator that models the processes of the sewage treatment plant and analyzes water quality data from the operation values ​​of the sewage treatment plant; an operation model learning unit which reinforces an operation model that outputs the operation values ​​based on the water quality data and operating costs analyzed by the plant model; an effluent water quality estimation unit which estimates future effluent water quality data based on at least past water quality data; a setting unit for setting target values ​​for water quality data and operating costs; and an operation unit which inputs the target values ​​set by the setting unit, the water quality data acquired by the acquisition unit, and the future effluent water quality data estimated by the effluent water quality estimation unit into the learned operation model and outputs the operation values ​​of the sewage treatment plant.

[0058] The operation value output device according to Embodiment 2 of the present disclosure further comprises an influent water quality estimation unit that estimates future influent water quality data based on at least past water quality data, the operation unit inputs a target value set by the setting unit, water quality data acquired by the acquisition unit, future influent water quality data estimated by the influent water quality estimation unit, and future effluent water quality data estimated by the effluent water quality estimation unit into the learned operation model and outputs the operation value of the sewage treatment plant.

[0059] A method for outputting operational values ​​according to aspect 3 of the present disclosure includes the steps of: acquiring water quality data in a sewage treatment plant; reinforcing a power operation model that outputs operational values ​​based on water quality data and operating costs analyzed by a plant model, which is a simulator that models the processes of the sewage treatment plant and analyzes water quality data from operational values ​​of the sewage treatment plant; estimating future effluent water quality data based on at least past water quality data; setting target values ​​for water quality data and operating costs; and inputting the set target values, the acquired water quality data, and the estimated future effluent water quality data into the trained power operation model to output operational values ​​of the sewage treatment plant.

[0060] The method for outputting operational values ​​according to aspect 4 of the present disclosure further includes, in aspect 3 above, a step of estimating future influent water quality data based on at least past water quality data, and in the step of outputting operational values ​​for the sewage treatment plant, inputting the set target value, the acquired water quality data, the estimated future influent water quality data, and the estimated future effluent water quality data into the trained operational model, and outputting operational values ​​for the sewage treatment plant.

[0061] The program according to aspect 5 of the present disclosure causes a computer to perform the following processes: acquiring water quality data in a sewage treatment plant; reinforcing a plant operation model, which is a simulator that models the processes of the sewage treatment plant and outputs the plant operation values ​​based on the water quality data analyzed by the plant model and the operating costs; estimating future effluent water quality data based on at least past water quality data; setting target values ​​for the water quality data and operating costs; and inputting the set target values, the acquired water quality data, and the estimated future effluent water quality data into the trained plant operation model and outputting the plant operation values.

[0062] The program according to aspect 6 of the present disclosure, in aspect 5 above, causes the computer to perform a process to estimate future influent water quality data based on at least past water quality data, and in the process of outputting the operating values ​​of the sewage treatment plant, inputs the set target value, the acquired water quality data, the estimated future influent water quality data, and the estimated future effluent water quality data into the trained operating model, and outputs the operating values ​​of the sewage treatment plant.

[0063] [Additional Notes] This disclosure is not limited to the embodiments described above, and various modifications are possible within the scope of the claims. Embodiments obtained by appropriately combining the technical means disclosed in different embodiments are also included in the technical scope of this disclosure. [Explanation of Symbols]

[0064] 1.1A Operation Value Output Device 2. Sewage treatment plant 10 Acquisition Department 11. Water Quality Data Storage Section 12. Operation Model Generation Unit 13,13A Water quality estimation section 14. Settings section 15,15A Operation Control Unit 100,100A wastewater treatment system 121 Plant Models 122 Driving Operation Model Learning Unit 123 Operation Model Storage Unit 131 Discharge Water Quality Estimation Model Learning Section 132 Discharge Water Quality Estimation Model Preservation Section 133 Discharge Water Quality Estimation Department 134 Inflow Water Quality Estimation Model Learning Unit 135 Inflow Water Quality Estimation Model Storage Section 136 Inflow Water Quality Estimation Department 141 Setting Input Section 142 Settings storage section

Claims

1. An acquisition unit that acquires water quality data in a sewage treatment plant, This simulator models the processes of the aforementioned sewage treatment plant, and includes a plant model that analyzes water quality data from the operating values ​​of the aforementioned sewage treatment plant. Based on the water quality data and operating costs analyzed by the plant model, the operation model learning unit reinforces the operation model that outputs the operation values, A discharge water quality estimation unit that estimates future discharge water quality data based on at least past water quality data, A setting unit for setting target values ​​for water quality data and operating costs, An operation value output device comprising: an operation unit that inputs a target value set by the setting unit, water quality data acquired by the acquisition unit, and future effluent water quality data estimated by the effluent water quality estimation unit into a learned operation model and outputs operation values ​​for the sewage treatment plant.

2. The aforementioned operating value output device further includes an inflow water quality estimation unit that estimates future inflow water quality data based on at least past water quality data, The operation unit inputs a target value set by the setting unit, water quality data acquired by the acquisition unit, future influent water quality data estimated by the influent water quality estimation unit, and future effluent water quality data estimated by the effluent water quality estimation unit into the learned operation model, and outputs an operation value for the sewage treatment plant, as described in claim 1.

3. Steps to acquire water quality data at a sewage treatment plant, A simulator that models the process of the aforementioned sewage treatment plant, comprising the step of reinforcing a control model that outputs the control values ​​based on water quality data analyzed by a plant model that analyzes water quality data from the operating values ​​of the aforementioned sewage treatment plant and the operating costs, The steps include, at least, estimating future discharge water quality data based on past water quality data, The steps include setting target values ​​for water quality data and operating costs, A method for outputting operating values, comprising the steps of inputting the set target value, the acquired water quality data, and the estimated future discharge water quality data into a trained operating model, and outputting operating values ​​for the sewage treatment plant.

4. The aforementioned operational value output method further includes the step of estimating future inflow water quality data based on at least past water quality data, The method for outputting operating values ​​according to claim 3, wherein in the step of outputting operating values ​​for the sewage treatment plant, the set target value, the acquired water quality data, the estimated future influent water quality data, and the estimated future effluent water quality data are input to the trained operating model to output operating values ​​for the sewage treatment plant.

5. On the computer, A process for acquiring water quality data at a sewage treatment plant, A simulator that models the process of the aforementioned sewage treatment plant, and a process that reinforces a control model that outputs the aforementioned control values ​​based on water quality data analyzed by a plant model that analyzes water quality data from the operating values ​​of the aforementioned sewage treatment plant and the operating costs, A process to estimate future discharge water quality data based on at least past water quality data, The process involves setting target values ​​for water quality data and operating costs, A program that inputs the set target value, the acquired water quality data, and the estimated future discharge water quality data into the trained operation model and performs the process of outputting the operation values ​​of the sewage treatment plant.

6. The computer is instructed to perform a process to estimate future inflow water quality data based on at least past water quality data. The program according to claim 5, which, in the process of outputting the operating values ​​of the sewage treatment plant, inputs the set target value, the acquired water quality data, the estimated future influent water quality data, and the estimated future effluent water quality data into the trained operating model, and outputs the operating values ​​of the sewage treatment plant.

Images