Water intake facility management system and water intake facility management method

Abstract

To detect overflow in a dam and a water-intake facility in advance to securely prevent generation of overflow electric power.SOLUTION: A water-intake facility management system managing a water-intake facility comprising a first screen for removing dust provided in a water-intake port for taking-in water used for water power generation from a dam: generates a first prediction model that is a machine learning model which learns time-series data of each of a water level of the dam, a water level of a water passage guiding water taken-in through the water-intake port to a water power plant, and an opening degree of a gate controlling an inflow amount of the water taken-in from the dam into the water passage as explanatory variables, and information exhibiting possibility of overflow from the dam as objective variables; and inputs newly acquired explanatory variables into the first prediction model, to output acquired objective variables. The first prediction model, further, outputs information exhibiting prediction time when overflow from the dam starts, as objective variables.SELECTED DRAWING: Figure 3

Description

[Technical field]

[0001] The present invention relates to a water intake equipment management system and a water intake equipment management method, and in particular to a technology for preventing a decrease in the power generation output of a hydroelectric power plant due to debris adhering to screens installed in the water intake equipment of the hydroelectric power plant. [Background technology]

[0002] Various technologies have been proposed to prevent a decrease in power generation output due to debris such as branches and leaves adhering to dust removal screens installed in equipment for taking water for hydroelectric power plants from dams (hereinafter referred to as "water intake equipment").

[0003] For example, Patent Document 1 describes a debris removal system in which, when water from a water source is guided through a water tank to the turbine of a generator of a hydroelectric power generation facility, debris adhering to a screen is removed by spraying gas or water from a spraying means provided at the bottom of the water tank toward the screen, and the spraying means is operated based on the water level in the water tank and the output of the generator.

[0004] Patent Document 2 describes an opening / closing control device for an intake gate that is designed to enable a constant amount of water to be taken in without causing a decrease in the amount of water taken in or breakage of an intrusion prevention member. The opening / closing control device monitors the presence or absence of a water level difference between the water level of the water flow and the water level downstream of the intrusion prevention member that prevents floating objects from entering the water conduit, closes the intake gate when there is a water level difference, and opens the intake gate a certain time after the intake gate is closed.

[0005] Patent Document 3 also describes a water supply system for a hydroelectric power plant. The water supply system includes a headrace that supplies water from a dam to a water tank, a penstock that sends water from the water tank to a generator, an intake that connects the water tank to the headrace, an outlet that connects to the penstock, and a debris trap screen that is disposed in the water tank to prevent debris that has entered the water tank from flowing into the outlet. The water tank is disposed adjacent to the upper region of the debris trap screen, and an overflow section is disposed above the full water level of the water tank. A spillway is further disposed connected to the overflow section and attached to the water tank, leading to a general river. [Prior art documents] [Patent documents]

[0006] [Patent Document 1] JP 2010-1673 A [Patent Document 2] JP 2007-77624 A [Patent Document 3] JP 2007-224682 A Summary of the Invention [Problem to be solved by the invention]

[0007] In the water intake facilities of hydroelectric power plants, the adhesion of dust to the screens installed in the intake and water tanks that constitute the water intake facilities leads to a decrease in power output, so the removal of dust attached to the screens is performed from time to time. However, currently, the manager of the water intake facility, etc., requests the contractor to perform dust removal work by telephone after confirming the start of overflow from the intake dam or water tank, so it is an issue to suppress the overflow power due to the time lag from when the overflow is confirmed to when the dust removal process is completed. In addition, if a dust remover is installed in the water intake or water tank, the dust remover will automatically start and remove dust from the screen when the water level difference before and after the screen exceeds a threshold value or when a predetermined time arrives. However, depending on the adhesion state of the dust to the screen, overflow may occur before the dust remover starts.

[0008] As a result, it is currently not possible to reliably prevent the occurrence of overflow power caused by debris adhering to the screen, and there is a need for a mechanism to predict overflow from intake dams and water tanks and reliably prevent the occurrence of overflow power.

[0009] The above-mentioned Patent Document 1 discloses a mechanism for removing dust adhering to the screen. Patent Document 2 discloses a mechanism for preventing a decrease in the amount of water intake and breakage of the intrusion prevention member. However, neither document discloses a mechanism for predicting the occurrence of overflow from the intake dam or water tank and reliably preventing the occurrence of overflow power. Patent Document 3 merely discloses a mechanism for reducing the frequency of dust removal from the dust capture screen placed in the water tank.

[0010] The present invention has been made in consideration of the above background, and aims to provide a water intake equipment management system and a water intake equipment management method that are capable of predicting the occurrence of overflow from water intake dams and water tanks and reliably preventing the occurrence of overflow power. [Means for solving the problem]

[0011] One of the present inventions for achieving the above-mentioned object is an information processing system that is configured using an information processing device having a processor and a storage device, and manages a water intake facility equipped with a first screen for removing debris, which is installed at an intake that takes in water to be used for hydroelectric power generation from a dam, and uses time series data for the water level of the dam, the water level of a waterway that leads the water taken from the intake to a hydroelectric power plant, and the opening degree of a gate that controls the amount of water taken from the dam into the waterway as explanatory variables, and generates a first prediction model, which is a machine learning model that has been trained using information indicating the possibility of overflow from the dam as a target variable, and determines and outputs the target variable by inputting the newly obtained explanatory variables into the first prediction model.

[0012] According to the present invention, by inputting the newly acquired explanatory variables (time series data of the water level of the dam, the water level of the waterway, and the gate opening) into the first prediction model, information indicating the possibility of overflow from the dam due to debris adhering to the first screen is output as the objective variable. Therefore, the manager of the water intake facility can detect the possibility of overflow from the dam in advance and take necessary measures, such as requesting debris removal from the first screen, at an early stage, thereby reliably preventing overflow from the dam. Therefore, overflow power can be reliably prevented.

[0013] Another aspect of the present invention is the water intake equipment management system, wherein the first prediction model further outputs, as the objective variable, information indicating a predicted time when overflow from the dam will begin.

[0014] According to the present invention, managers and others can know the predicted time when overflow from the dam will begin, and can use the predicted time as a guide to efficiently take necessary measures, such as requesting dust removal from the first screen, at the appropriate time.

[0015] Another aspect of the present invention is the above-mentioned water intake equipment management system, wherein the water intake equipment further includes a first dust remover that automatically removes dust adhering to the first screen, and the first predictive model further outputs information indicating whether or not the first dust remover needs to be started as the objective variable.

[0016] According to the present invention, the first dust remover can be started at an appropriate time based on information indicating whether or not the first dust remover needs to be started, and the first dust remover can be used efficiently to reliably prevent overflow from the dam.

[0017] Another aspect of the present invention is the above-mentioned water intake equipment management system, which is communicatively connected to the first dust remover, and when the information output as the objective variable indicates a need to start the first dust remover, sends a control instruction to the first dust remover to start the first dust remover.

[0018] Another aspect of the present invention is the above-mentioned water intake equipment management system, wherein the water intake equipment further includes a second screen for removing debris provided in a water tank that stores water guided by the water channel and supplies the water to the hydraulic pipeline of the hydroelectric power plant, and a second prediction model is generated, which is a machine learning model that uses time series data on the water level of the water tank and the power output of the hydroelectric power plant as explanatory variables and information indicating the possibility of overflow from the water tank as a target variable, and the newly acquired explanatory variables are input into the second prediction model to determine and output the target variable.

[0019] According to the present invention, by inputting the newly acquired explanatory variables (time series data of the water level of the water tank and the power output of the hydroelectric power plant) into the second prediction model, information indicating the possibility of overflow from the water tank due to debris adhering to the second screen is output as the objective variable. Therefore, the manager of the water intake facility can detect the possibility of overflow from the water tank in advance and take necessary measures such as requesting dust removal from the second screen at an early stage, thereby reliably preventing overflow from the water tank. Therefore, overflow power can be reliably prevented.

[0020] Another aspect of the present invention is the water intake equipment management system, wherein the second prediction model further outputs, as the objective variable, information indicating a predicted time when overflow from the water tank will start.

[0021] According to the present invention, managers and others can know the predicted time when overflow from the tank will begin, and can use the predicted time as a guide to efficiently take necessary measures, such as requesting dust removal from the second screen, at the appropriate time.

[0022] Another aspect of the present invention is the above-mentioned water intake equipment management system, wherein the water intake equipment further includes a second dust remover that automatically removes dust adhering to the second screen, and the second predictive model further outputs information indicating whether or not the second dust remover needs to be started as the objective variable.

[0023] According to the present invention, the second dust remover can be started at an appropriate time based on information indicating whether or not the second dust remover needs to be started, and the second dust remover can be used efficiently to reliably prevent overflow from the dam.

[0024] Another aspect of the present invention is the above-mentioned water intake equipment management system, which is communicatively connected to the second dust remover, and when the information output as the objective variable indicates a need to start the second dust remover, sends a control instruction to the second dust remover to start the second dust remover.

[0025] Other problems and solutions disclosed in the present application will be made clear in the section of the description of the invention and the drawings. Effect of the Invention

[0026] According to the present invention, the occurrence of overflow from a water intake dam or a water tank can be predicted, and the occurrence of overflow power can be reliably prevented. [Brief description of the drawings]

[0027] [Figure 1A] FIG. 2 is a diagram showing an example of the configuration of the water intake facility around the water intake. [Figure 1B] FIG. 2 is a diagram showing an example of the configuration of a water intake facility in the vicinity of a power plant. [Diagram 2] FIG. 2 is a diagram showing a schematic configuration of a water intake facility management system. [Diagram 3] 2 is a diagram illustrating the main functions of the water intake equipment management device. FIG. [Figure 4] FIG. 2 is a diagram illustrating main functions of the monitoring control device. [Diagram 5] 1 is an example of a hardware configuration of an information processing device used to realize a water intake facility management device or a monitoring control device. [Figure 6A] This is an example of training data (dam side). [Figure 6B] This is an example of training data (power plant side). [Figure 7A] This is an example of a predictive model (dam side) (neural network structure). [Figure 7B] This is an example of a predictive model (power plant side) (neural network structure). [Figure 8A] 13 is a flowchart illustrating a prediction model (dam side) learning process. [Figure 8B] 13 is a flowchart illustrating a prediction model (power plant side) learning process. [Figure 9A] 13 is a flowchart illustrating an initial speed process (dam side). [Figure 9B] 13 is a flowchart illustrating the initial speed processing (power plant side). [Figure 10A] This is an example of a prediction result display screen (dam side). [Figure 10B] This is an example of a prediction result display screen (power plant side). DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0028] Hereinafter, an embodiment will be described with reference to the drawings. In the following description, the same or similar components will be denoted by the same reference numerals, and the description thereof will be omitted.

[0029] 1A and 1B show the configuration of a water intake facility 2 of a hydroelectric power plant (hereinafter referred to as a "power plant 31"), which will be described as one embodiment of the present invention. Fig. 1A shows an example of the configuration of the water intake facility 2 around an intake port 21 that takes in water for use from a dam, and Fig. 1B shows an example of the configuration of the water intake facility 2 around a power plant 31.

[0030] As shown in FIG. 1A, around the water intake 21, a first screen 22, a gate 24, a gate operator 25, a first dust remover 28, a dam water level gauge 26, and a waterway water level gauge 27 are provided.

[0031] Of these, the first screen 22 prevents debris D, such as branches and leaves, contained in the water taken in from the dam 20 from flowing into the waterway 23. The gate 24 adjusts the amount of water flowing from the dam 20 into the waterway 23. The gate operator 25 controls the opening and closing of the gate 24. The first dust remover 28 operates in response to a control signal from the monitoring and control device 200, and automatically removes debris D adhering to the first screen 22.

[0032] The dam water level gauge 26 measures the water level of the dam 20 in real time and transmits the measurement value to the monitoring control device 200. The waterway water level gauge 27 measures the water level of the waterway 23 in real time and transmits the measurement value to the monitoring control device 200.

[0033] In addition, in the configuration shown in the same figure, overflow from the dam 20 caused by debris D adhering to the first screen 22 occurs when the water level of the waterway 23 drops due to the debris D adhering to the first screen 22, and the adhesion of the debris D progresses, causing the water level of the dam to continue to rise.

[0034] As shown in FIG. 1B, around the power plant 31, there are provided a water tank 32, a spillway 33, a hydraulic pipeline 34, a second screen 35, a second dust collector 39, a water tank water level gauge 36, an upper pipeline water level gauge 37, and a power output meter 38.

[0035] Of these, the water tank 32 stores water from the water channel 23. The spillway 33 is provided next to the water tank 32 and drains water that overflows from the water tank 32. The penstock 34 guides the water stored in the water tank 32 to the power plant 31. The second screen 35 prevents debris D contained in the water stored in the water tank 32 from flowing into the penstock 34. The second dust remover 39 operates in response to a control signal from the monitoring and control device 200, and automatically removes debris D adhering to the second screen 35.

[0036] The water tank water level gauge 36 measures the water level of the water tank 32 in real time and transmits the measurement value to the monitoring and control device 200 described below. The upper pipe water level gauge 37 measures the water level of the upper pipe (the connection between the water tank 32 and the hydraulic pipe 34 behind the second screen 35) in real time and transmits the measurement value to the monitoring and control device 200. The power generation output meter 38 measures the power generation output of the power plant 31 and transmits the measurement value to the monitoring and control device 200.

[0037] In the configuration shown in the same figure, overflow of the water tank 32 (overflow from the water tank 32 to the spillway 33) caused by debris D adhering to the second screen 35 occurs when the water level in the water storage area behind the second screen 35 (the upper water storage area of ​​the water pressure pipe 34) drops due to the debris D adhering to the second screen 35, and the adhesion of the debris D progresses, causing the water level in the water tank 32 to continue to rise.

[0038] Fig. 2 shows a schematic configuration of a water intake equipment management system 1 of this embodiment, which is applied to the water intake equipment 2 shown in Fig. 1A and Fig. 1B. As shown in the figure, the water intake equipment management system 1 includes a water intake equipment management device 100 and a monitoring control device 200.

[0039] The water intake equipment management device 100 and the monitoring control device 200 are connected to each other so as to be able to communicate with each other by wireless or wired communication via a communication infrastructure 5. The monitoring control device 200 is also connected to the gate opening / closing device 25, the dam water level gauge 26, the waterway water level gauge 27, the first dust remover 28, the water tank water level gauge 36, the upper pipe water level gauge 37, the power generation output gauge 38, and the second dust remover 39 so as to be able to communicate with each other by wireless or wired communication via a communication infrastructure 6. The communication infrastructures 5 and 6 are configured using, for example, a communication infrastructure such as a LAN (Local Area Network), a WAN (Wide Area Network), a dedicated line, the Internet, a 920 Hz band communication network, or a PLC (Power Line Communication).

[0040] The water intake equipment management device 100 monitors the state of the water intake equipment 2 based on the measurement values ​​and control information sent from each device (dam water level gauge 26, water channel water level gauge 27, gate operator 25, first dust remover 28, water tank water level gauge 36, upper pipe water level gauge 37, power generation output meter 38, second dust remover 39) installed in the water intake equipment 2. In addition, the water intake equipment management device 100 transmits control instructions for each component of the water intake equipment 2 (for example, control instructions for the gate operator 25, control instructions for the first dust remover 28, control instructions for the second dust remover 39, etc.) to the monitoring control device 200 via the communication infrastructure 5.

[0041] Fig. 3 is a diagram for explaining main functions of the water intake equipment management device 100 shown in Fig. 2. As shown in the figure, the water intake equipment management device 100 has the functions of a memory unit 110, a measurement control information acquisition management unit 120, a learning data generation unit 125, a model learning unit 130, a prediction processing unit 135, a prediction result presentation unit 140, and a dust remover control unit 145.

[0042] Of the above functions, the memory unit 110 stores measurement control information (dam side) 111a, measurement control information (power plant side) 111b, learning data (dam side) 112a, learning data (power plant side) 112b, prediction model (dam side) 113a (first prediction model), prediction model (power plant side) 113b (second prediction model), prediction result (dam side) 114a, and prediction result (power plant side) 114b.

[0043] The measurement control information (dam side) 111a includes each measurement value of the dam water level gauge 26 and the water channel water level gauge 27, and control information (control history) of the gate opening / closing device 25 and the first dust remover 28, which the water intake equipment management device 100 receives from the monitoring control device 200 via the communication infrastructure 5. The measurement control information (power plant side) 111b includes each measurement value of the water tank water level gauge 36, the upper pipe water level gauge 37, and the power generation output meter 38, and control information (control history) of the second dust remover 39, which the water intake equipment management device 100 receives from the monitoring control device 200 via the communication infrastructure 5.

[0044] The learning data (dam side) 112a is learning data (teacher data) used to learn a prediction model (dam side) 113a, which is generated based on the measurement control information (dam side) 111a. The learning data (power plant side) 112b is learning data (teacher data) used to learn a prediction model (power plant side) 113b, which is generated based on the measurement control information (power plant side) 111b.

[0045] The prediction model (dam side) 113a is a machine learning model that predicts future states around the intake 21 (presence or absence of overflow from the dam 20, the start time of overflow from the dam 20, whether or not control of the gate opening / closing device 25 is required and the control amount, whether or not control of the first dust remover 28 is required and the control amount, etc.) based on the measurement control information (dam side) 111a. The prediction model (power plant side) 113b is a machine learning model that predicts future states around the power plant 31 (presence or absence of overflow from the water tank 32, the start time of overflow from the water tank 32, whether or not control of the second dust remover 39 is required and the control amount, etc.) based on the measurement control information (power plant side) 111b.

[0046] The specific implementation method of the prediction model (dam side) 113a and the prediction model (power plant side) 113b is not necessarily limited, but may be implemented, for example, by a deep neural network (DNN), gradient boosting (Gradient Boosting Decision Tree (GBDT)), logistic regression, k-means, etc. In this embodiment, a case in which both the prediction model (dam side) 113a and the prediction model (power plant side) 113b are implemented by DNN will be described as an example. The substance of the prediction model (dam side) 113a and the prediction model (power plant side) 113b is expressed, for example, by a determinant, a formula, a vector, etc., including adjustable parameters.

[0047] The prediction result (dam side) 114a is a result predicted by the prediction model (dam side) 113a, and includes information indicating the future state of the area around the water intake 21. The prediction result (power plant side) 114b is a result predicted by the prediction model (power plant side) 113b, and includes information indicating the future state of the area around the power plant 31.

[0048] The measurement control information acquisition and management unit 120 shown in the same figure receives measurement values ​​and control information sent from each device installed in the water intake facility 2 (dam water level gauge 26, water channel water level gauge 27, gate opening / closing device 25, first dust remover 28, water tank water level gauge 36, upper pipeline water level gauge 37, power generation output meter 38, second dust remover 39), and manages the measurement values ​​and control information received from devices installed around the dam as measurement control information (dam side) 111a, and the measurement values ​​and control information received from devices installed around the hydroelectric power plant as measurement control information (power plant side) 111b.

[0049] In this embodiment, the measurement control information acquisition management unit 120 acquires the measurement values ​​and control information sent from each device at a predetermined time interval (for example, every minute, every 10 minutes, every 30 minutes, etc.) and manages them as time-series data. The measurement control information acquisition management unit 120 manages both the time-series data newly acquired from each device and the time-series data previously acquired from each device as measurement control information (dam side) 111a and measurement control information (power plant side) 111b.

[0050] The learning data generation unit 125 generates learning data (dam side) 112a, which is learning data (teacher data) used for learning a prediction model (dam side) 113a, based on the measurement control information (dam side) 111a. Also, the learning data generation unit 125 generates learning data (power plant side) 112b, which is learning data (teacher data) used for learning a prediction model (power plant side) 113b, based on the measurement control information (power plant side) 111b.

[0051] The learning data generating unit 125 may present the time series data of the measurement values ​​in a predetermined period (predetermined time, half a day, a full day, a week, etc., hereinafter referred to as "time interval") acquired in the past to the user as explanatory variables (feature amounts), and accept from the user the setting of a label that is a target variable (correct answer data) for the measurement values. The above-mentioned time interval may be empirically set based on, for example, the time series changes of the measurement values ​​leading up to overflow when overflow occurred in the past.

[0052] The model learning unit 130 uses the learning data (dam side) 112a to learn a prediction model (dam side) 113a. Also, the model learning unit 130 uses the learning data (power plant side) 112b to learn a prediction model (power plant side) 113b.

[0053] Specifically, the model learning unit 130 inputs the learning data (learning data (dam side) 112a, learning data (power plant side) 112b) into the corresponding prediction model (prediction model (dam side) 113a, prediction model (power plant side) 113b) as explanatory variables, and learns the prediction model by adjusting the parameters defining the prediction model based on the difference between the result output by the prediction model and the objective variable in the learning data, for example, by a method such as backpropagation.

[0054] The model learning unit 130 learns, for example, the patterns of time-series changes in explanatory variables (learning data (dam side) 112a and learning data (power plant side) 112b) leading up to overflow as features, thereby learning the prediction model (dam side) 113a and the prediction model (power plant side) 113b.

[0055] The prediction processing unit 135 inputs the explanatory variables (measurement control information (dam side) 111a) for a newly acquired specified time interval into the prediction model (dam side) 113a to obtain objective variables (information indicating the possibility of overflow from the dam 20, information indicating the predicted time when overflow from the dam 20 will begin, information indicating whether the first dust remover 28 needs to be started, etc.), and manages the obtained objective variables as prediction results (dam side) 114a.

[0056] In addition, the prediction processing unit 135 inputs the explanatory variables (measurement control information (power plant side) 111b) for a newly acquired specified time interval into the prediction model (power plant side) 113b to obtain objective variables (information indicating the possibility of overflow from the water tank, information indicating the predicted time when overflow from the water tank will start, information indicating whether the second dust remover 39 needs to be started, etc.), and manages the obtained objective variables as prediction results (power plant side) 114b.

[0057] The prediction result presentation unit 140 generates a screen (prediction result presentation screen (dam side) 1000a and prediction result presentation screen (power plant side) 1000b described below) that lists the contents of the prediction result (dam side) 114a and the prediction result (power plant side) 114b, and presents (displays) the generated screen to a user such as the manager of the water intake facility 2 via a user interface.

[0058] The dust remover control unit 145 generates control instructions for the gate opening / closing unit 25, the first dust remover 28, and the second dust remover 39 based on the objective variables obtained by the prediction processing unit 135, and transmits the generated control instructions to the monitoring control device 200.

[0059] 4 is a diagram for explaining main functions of the monitoring control device 200. As shown in the figure, the monitoring control device 200 has the functions of a storage unit 210, a measurement value acquisition and transfer unit 220, a control instruction receiving unit 225, a gate opening / closing device control unit 230, a first dust remover control unit 235, and a second dust remover control unit 240.

[0060] Of the above functions, the storage unit 210 stores measurement value information 211 and control instruction information 212. Details of this information will be described later.

[0061] The measurement value acquisition and transfer unit 220 receives measurement values ​​sent from each device installed in the water intake equipment 2 (dam water level gauge 26, water channel water level gauge 27, tank water level gauge 36, upper pipeline water level gauge 37, power generation output meter 38), manages the received measurement values ​​as measurement value information 211, and transmits (transfers) the measurement values ​​to the water intake equipment management device 100.

[0062] The control instruction receiving unit 225 receives control instructions (control instructions for the gate opening / closing machine 25, control instructions for the first dust remover 28, and control instructions for the second dust remover 39) sent from the water intake equipment management device 100, and manages the received control instructions as control instruction information 212.

[0063] The gate operator control unit 230 transmits a control signal to the gate operator 25 in response to a control instruction sent from the water intake equipment management device 100 to control the opening and closing of the gate 24 .

[0064] The first dust remover control unit 235 transmits a control signal to the first dust remover 28 in response to a control instruction sent from the water intake equipment management device 100 to control the operation of the first dust remover 28 .

[0065] The second dust remover control unit 240 transmits a control signal to the second dust remover 39 in response to a control instruction sent from the water intake equipment management device 100 to control the operation of the second dust remover 39 .

[0066] 5 is an example of a hardware configuration of an information processing device used to realize the water intake facility management device 100 and the monitoring control device 200. The illustrated information processing device 10 includes a processor 11, a main memory device 12, an auxiliary memory device 13, an input device 14, an output device 15, and a communication device 16. Specific examples of the information processing device 10 include personal computers, office computers, various server devices, general-purpose machines, and the like. The information processing device 10 may be realized in whole or in part using virtual information processing resources provided using virtualization technology, process space separation technology, and the like, such as a virtual server provided by a cloud system. The water intake facility management device 100 and the monitoring control device 200 may be realized using a plurality of information processing devices 10 connected to each other so as to be able to communicate with each other.

[0067] In the figure, the processor 11 is configured using, for example, a CPU (Central Processing Unit), an MPU (Micro Processing Unit), a GPU (Graphics Processing Unit), an FPGA (Field Programmable Gate Array), an ASIC (Application Specific Integrated Circuit), an AI (Artificial Intelligence) chip, etc.

[0068] The main memory device 12 is a device that stores programs and data, and is, for example, a Read Only Memory (ROM), a Random Access Memory (RAM), or a non-volatile memory (Non Volatile RAM (NVRAM)).

[0069] The auxiliary storage device 13 is, for example, an SSD (Solid State Drive), a hard disk drive, an optical storage device (CD (Compact Disc), DVD (Digital Versatile Disc), etc.), a storage system, an IC card, a read / write device for a recording medium such as an SD card or an optical recording medium, a storage area of ​​a cloud server, etc. Programs and data can be read into the auxiliary storage device 13 via a recording medium reader or a communication device 16. The programs and data stored (memorized) in the auxiliary storage device 13 are read into the main storage device 12 as needed.

[0070] The input device 14 is an interface that accepts input from the outside, and is, for example, a keyboard, a mouse, a touch panel, a card reader, a pen-input tablet, a voice input device, or the like.

[0071] The output device 15 is an interface that outputs various information such as the progress of processing and the results of processing. The output device 15 is, for example, a display device (LCD (Liquid Crystal Display), a graphic card, etc.) that visualizes the various information described above, a device that converts the various information described above into voice (voice output device (speaker, etc.)), or a device that converts the various information described above into text (printer, etc.). Note that, for example, the information processing device 10 may be configured to input and output information to and from other devices via the communication device 16.

[0072] The input device 14 and the output device 15 constitute a user interface for receiving information from the user and presenting information.

[0073] The communication device 16 is a device that realizes communication (wired communication or wireless communication) with other devices via the communication infrastructure 5 or the communication infrastructure 6, and is configured using, for example, a NIC (Network Interface Card), a wireless communication module, etc.

[0074] The information processing device 10 may be implemented with, for example, an operating system, a file system, a DBMS (DataBase Management System) (relational database, NoSQL, etc.), a KVS (Key-Value Store), etc.

[0075] The functions of the water intake equipment management device 100 and the monitoring control device 200 are realized by the processor 11 of the information processing device 10 reading and executing programs stored in the main memory device 12. The water intake equipment management device 100 and the monitoring control device 200 store the various information (data) described above, for example, as tables in a database or files managed by a file system.

[0076] 6A shows an example of the learning data (dam side) 112a. The learning data (dam side) 112a shown in the example is composed of a plurality of records having the following items (fields): measurement date and time 1121a, dam water level 1122a, waterway water level 1123a, gate opening 1124a, gate operation method 1125a, dam overflow presence / absence 1126a, and first dust collector start / absence 1127a. One record of the learning data (dam side) 112a corresponds to a measurement value measured at a certain date and time. Among the above items, the measurement date and time 1121a, dam water level 1122a, gate opening 1124a, and gate operation method 1125a constitute explanatory variables in the prediction model (dam side) 113a. Among the above items, the dam overflow presence / absence 1126a and the first dust remover start-up presence / absence 1127a configure the objective variables in the prediction model (dam side) 113a.

[0077] Of the above items, the measurement date and time 1121a stores the date and time when the measurement value of the record was measured. The dam water level 1122a stores the water level of the dam 20 measured at the date and time. The waterway water level 1123a stores the water level of the waterway 23 measured at the date and time. The gate opening 1124a stores the opening of the gate 24 at the date and time. The gate operation method 1125a stores information indicating the operation state of the gate 24 applied at the date and time (constant dam water level (when operation is performed to maintain the dam water level at a predetermined set value), automatic upper limit (when the gate 24 has reached a preset upper limit opening), etc.). The dam overflow presence / absence 1126a stores information indicating whether overflow from the dam 20 has occurred at the date and time ("Yes" if overflow has occurred, and "No" if overflow has not occurred). The first dust remover activation status 1127a stores information indicating whether the first dust remover 28 was activated at the relevant date and time ("Yes" if activated, and "No" if not activated).

[0078] FIG. 6B shows an example of the learning data (power plant side) 112b. The learning data (power plant side) 112b shown in the example is composed of a plurality of records having the items (fields) of measurement date and time 1121b, water tank water level 1122b, power output 1123b, water tank overflow presence / absence 1124b, and second dust remover start / absence 1125b. One record of the learning data (power plant side) 112b corresponds to a measurement value measured at a certain date and time. Among the above items, the measurement date and time 1121b, water tank water level 1122b, and power output 1123b constitute explanatory variables in the prediction model (power plant side) 113b. Among the above items, the water tank overflow presence / absence 1124b and second dust remover start / absence 1125b constitute objective variables in the prediction model (power plant side) 113b.

[0079] Of the above items, the measurement date and time 1121b stores the date and time when the measurement value of the record was measured. The water tank water level 1122b stores the water level of the water tank 32 measured at the date and time. The power generation output 1123b stores the output of the power plant 31 (generator) measured at the date and time. The gate opening 1124a stores the opening of the gate 24 at the date and time. The water tank overflow presence / absence 1124b stores information indicating whether or not overflow from the water tank 32 is occurring at the date and time ("Yes" if overflow is occurring, and "No" if not). The second dust remover start / stop status 1125b stores information indicating whether or not the second dust remover 39 was started at the date and time ("Yes" if started, and "No" if not started).

[0080] FIG. 7A shows an example (structure of a neural network) of the prediction model (dam side) 113a. As shown in the figure, time series data of measurement values ​​(water level of the dam 20, water level of the waterway 23, opening degree of the gate 24, etc.) in a predetermined time interval acquired from the measurement control information (dam side) 111a is input to the input layer 711a of the prediction model (dam side) 113a. The intermediate layer 712a includes one or more hidden layers consisting of one or more nodes including parameters adjusted by learning. Based on the explanatory variables given to the input layer 711a, the intermediate layer 712a obtains one or more objective variables of the output layer 713a (information indicating the possibility of overflow from the dam 20 (probability of overflow from the dam 20, etc.), predicted time when overflow from the dam will start (time until overflow, etc.), information indicating whether or not the first dust remover 28 needs to be started (probability of the first dust remover 28 needing to be started, etc.).

[0081] FIG. 7B shows an example (structure of a neural network) of the prediction model (power plant side) 113b. As shown in the figure, time series data of measurement values ​​(water level of the water tank 32, power output of the power plant 31, etc.) in a predetermined time interval acquired from the measurement control information (power plant side) 111b is input to the input layer 711b of the prediction model (power plant side) 113b. The intermediate layer 712b includes one or more hidden layers consisting of one or more nodes including parameters adjusted by learning. Based on the explanatory variables given to the input layer 711b, the intermediate layer 712b calculates one or more objective variables of the output layer 713b (information indicating the possibility of overflow from the water tank 32 (probability of overflow from the water tank 32, etc.), predicted time when overflow from the water tank 32 will start (time until overflow, etc.), information indicating whether or not the second dust remover 39 needs to be started (probability of needing to start the second dust remover 39), etc.).

[0082] 8A is a flowchart for explaining the process (hereinafter referred to as "prediction model (dam side) learning process S800a") that the water intake equipment management device 100 performs when generating learning data (dam side) 112a and learning the prediction model (dam side) 113a using the generated learning data (dam side) 112a. The timing for executing the prediction model (dam side) learning process S800a is not necessarily limited, but for example, the water intake equipment management device 100 executes the prediction model (dam side) learning process S800a when new learning data (dam side) 112a is generated or when an execution instruction is received from a user via a user interface. The prediction model (dam side) learning process S800a will be explained below with reference to the same figure.

[0083] First, the learning data generating unit 125 of the water intake equipment management device 100 generates learning data (dam side) 112a based on the measurement values ​​(explanatory variables, objective variables) of the measurement control information (dam side) 111a (S811a).

[0084] Next, the model learning unit 130 of the water intake equipment management device 100 learns the prediction model (dam side) 113a based on the generated learning data (dam side) 112a (S812a). The model learning unit 130 may further verify the prediction accuracy of the learned prediction model (dam side) 113a, for example. In that case, for example, the learning data (dam side) 112a is classified in advance into data for learning and data for verification, and the prediction model (dam side) 113a is learned using the data for learning, and the prediction model (dam side) 113a is verified using the data for verification.

[0085] 8B is a flowchart for explaining the process (hereinafter referred to as "prediction model (power plant side) learning process S800b") that the water intake equipment management device 100 performs when generating learning data (power plant side) 112b and learning the prediction model (power plant side) 113b using the generated learning data (power plant side) 112b. The timing for executing the prediction model (power plant side) learning process S800b is not necessarily limited. For example, the water intake equipment management device 100 executes the prediction model (power plant side) learning process S800b when new learning data (power plant side) 112b is generated or when an execution instruction is received from a user via a user interface. The prediction model (power plant side) learning process S800b will be explained below with reference to FIG.

[0086] First, the learning data generating unit 125 of the water intake equipment management device 100 generates learning data (power plant side) 112b based on the measurement values ​​(explanatory variables, objective variables) of the measurement control information (power plant side) 111b (S811b).

[0087] Next, the model learning unit 130 of the water intake equipment management device 100 learns the prediction model (power plant side) 113b based on the generated learning data (power plant side) 112b (S812b). The model learning unit 130 may further verify the prediction accuracy of the learned prediction model (power plant side) 113b, for example. In that case, for example, the learning data (power plant side) 112b is classified in advance into data for learning and data for verification, and the prediction model (power plant side) 113b is learned using the data for learning, and the prediction model (power plant side) 113b is verified using the data for verification.

[0088] 9A is a flowchart for explaining a process (hereinafter referred to as "prediction process (dam side) S900a") in which the water intake equipment management device 100 obtains an objective variable by inputting a newly acquired explanatory variable (for example, time series data of the measurement value of the measurement control information (dam side) 111a in a predetermined time interval newly acquired) into the prediction model (dam side) 113a, presents the obtained objective variable to a user, and controls the first dust remover 28. The prediction process (dam side) S900 is executed at a preset timing (when newly acquired measurement control information (dam side) 111a, periodically, at a scheduled date and time, when an execution instruction is received from a user, etc.). The prediction process (dam side) S900a will be explained below with reference to the same figure.

[0089] First, the prediction processing unit 135 acquires the newly acquired explanatory variables from the measurement control information (dam side) 111a (S911a).

[0090] Next, the prediction processing unit 135 inputs the obtained explanatory variables to the prediction model (dam side) 113a, and obtains the output of the prediction model (dam side) 113a as the prediction result (dam side) 114a (S912a).

[0091] Next, the prediction result presentation unit 140 generates a screen (hereinafter referred to as the "prediction result presentation screen (dam side) 1000a") that displays the contents of the prediction result (dam side) 114a, and outputs (displays) the generated prediction result presentation screen (dam side) 1000a on the output device 15 (display device) (S913a).

[0092] Next, the dust remover control unit 145 judges whether or not to start the first dust remover 28. When starting the first dust remover 28, the dust remover control unit 145 transmits a start instruction of the first dust remover 28 to the monitoring control device 200 to start the first dust remover 28 and remove the dust D adhering to the first screen 22 (S914a). Note that the dust remover control unit 145 starts the first dust remover 28 when, for example, a user who has viewed the prediction result presentation screen (dam side) 1000a performs an operation to request the start of the first dust remover 28 via a user interface. In addition, the dust remover control unit 145 judges to start the first dust remover 28 when, for example, the overflow start time of the dam 20 specified from the prediction result (dam side) 114a is within a predetermined time from the present.

[0093] 9B is a flowchart explaining a process (hereinafter referred to as "prediction process (power plant side) S900b") in which the water intake equipment management device 100 obtains an objective variable by inputting a newly acquired explanatory variable (for example, time series data of the measurement value of the measurement control information (power plant side) 111b in a predetermined time interval newly acquired) into the prediction model (power plant side) 113b, and presents the obtained objective variable to a user and controls the second dust remover 39. The prediction process (power plant side) S900b is executed at a preset timing (when newly acquired measurement control information (power plant side) 111b, periodically, at a scheduled date and time, when an execution instruction is received from a user, etc.). The prediction process (power plant side) S900b will be explained below with reference to the same figure.

[0094] First, the prediction processing unit 135 acquires the newly acquired explanatory variables from the measurement control information (power plant side) 111b (S911a).

[0095] Next, the prediction processing unit 135 inputs the acquired explanatory variables to the prediction model (power plant side) 113b, and acquires the output of the prediction model (power plant side) 113b as the prediction result (power plant side) 114b (S912b).

[0096] Next, the prediction result presentation unit 140 generates a screen (hereinafter referred to as the "prediction result presentation screen (power plant side) 1000b") that displays the contents of the prediction result (power plant side) 114b, and outputs (displays) the generated prediction result presentation screen (power plant side) 1000b on the output device 15 (display device) (S913b).

[0097] Next, the dust remover control unit 145 judges whether or not to start the second dust remover 39. When starting the second dust remover 39, the dust remover control unit 145 transmits an instruction to start the second dust remover 39 to the monitoring control device 200 to start the second dust remover 39 and remove the dust D adhering to the second screen 35 (S914b). The dust remover control unit 145 starts the second dust remover 39 when, for example, a user who has viewed the prediction result presentation screen (power plant side) 1000b performs an operation to request the start of the second dust remover 39 via a user interface. In addition, the dust remover control unit 145 judges to start the second dust remover 39 when, for example, the overflow start time of the water tank 32 specified from the prediction result (power plant side) 114b is within a predetermined time from the present.

[0098] FIG. 10A shows an example of a prediction result display screen (dam side) 1000a. The illustrated prediction result display screen (dam side) 1000a has a display field 1010a for the time interval of the prediction target and a display field 1020a for the prediction result. The display field 1010a for the time interval of the prediction target displays information indicating the time interval of the time series data of the measurement value of the prediction target. The display field 1020a for the prediction result displays the contents of the prediction result (dam side) 114a obtained in S912a of FIG. 9A. As shown in FIG. 10A, the display field 1020a for the prediction result shows the probability 1011a of the dam overflowing, the time 1012a until the dam overflows, the necessity 1013a of starting the first dust remover, and the like, which are objective variables.

[0099] FIG. 10B shows an example of a prediction result display screen (power plant side) 1000b. The illustrated prediction result display screen (power plant side) 1000b has a display field 1010b for the time interval of the prediction target and a display field 1020b for the prediction result. The display field 1010b for the time interval of the prediction target displays information indicating the time interval of the time series data of the measurement value of the prediction target. The display field 1020b for the prediction result displays the contents of the prediction result (power plant side) 114b obtained in S912b of FIG. 9B. As shown in FIG. 10B, the display field 1020b for the prediction result displays the objective variables, such as the probability 1011b that the water tank 32 will overflow, the time 1012b until the water tank 32 overflows, and the necessity 1013b of starting the second dust remover.

[0100] As described above in detail, according to the water intake facility management system 1 of this embodiment, by inputting the newly acquired explanatory variables (time series data of the water level of the dam, the water level of the waterway, and the opening degree of the gate) into the prediction model (dam side) 113a, information indicating the possibility of overflow from the dam 20 due to the adhesion of the debris D to the first screen 22, the predicted time when the overflow from the dam 20 will start, and information indicating the necessity of starting the first dust remover 28 are output as objective variables. Therefore, the manager of the water intake facility 2 can know in advance the possibility of overflow from the dam 20, and can request the removal of the debris attached to the first screen 22 at an early stage, thereby reliably preventing the occurrence of overflow from the dam 20. Therefore, the occurrence of overflow power in the power plant 31 can be reliably prevented.

[0101] Moreover, according to the water intake facility management system 1 of this embodiment, by inputting the newly acquired explanatory variables (time series data of the water level of the water tank 32 and the power output of the power plant 31) into the prediction model (power plant side) 113b, information indicating the possibility of overflow from the water tank 32 due to the adhesion of debris D to the second screen 35 is output as a target variable. Therefore, the manager of the water intake facility 2, etc. can know in advance the possibility of overflow from the water tank 32 occurring, and can request dust removal from the second screen 35 at an early stage, thereby reliably preventing overflow from the water tank 32. Therefore, overflow power in the power plant 31 can be reliably prevented.

[0102] Although the embodiment of the present invention has been described above in detail, the above description is for the purpose of facilitating understanding of the present invention and does not limit the present invention. The present invention may be modified or improved without departing from the spirit thereof, and the present invention naturally includes equivalents thereof. For example, the above embodiment has been described in detail to explain the present invention in an easily understandable manner, and is not necessarily limited to those having all of the configurations described. In addition, it is possible to add, delete, or replace part of the configuration of the above embodiment with other configurations.

[0103] For example, in the above embodiments, separate prediction models were generated for the dam side and the power plant side to predict the state of the dam side and the power plant side, but it is also possible to generate a prediction model that predicts the state of the dam side and the power plant side as the objective variable by using explanatory variables that combine explanatory variables (measured values) acquired on the dam side and explanatory variables (measured values) acquired on the power plant side as input, and use the prediction model to predict the future state of the dam side and the power plant side. [Explanation of symbols]

[0104] 1 Water intake facility management system, 2 Water intake facility, 5 Communication infrastructure, 6 Communication infrastructure, 21 Water intake, 22 First screen, 24 Gate, 25 Gate opening / closing mechanism, 26 Dam water level gauge, 27 Waterway water level gauge, 28 First dust remover, 32 Water tank, 33 Spillway, 34 Water pressure pipe, 35 Second screen, 36 Water tank water level gauge, 37 Pipe upper water level gauge, 38 Power generation output gauge, 39 Second dust remover, 100 Water intake facility management device, 110 Memory unit, 111a Measurement control information (dam side), 111b Measurement control information (power plant side), 112a Learning data (dam side), 112b Learning data (power plant side), 113a Prediction model (dam side), 113b Prediction model (power plant side), 114a Prediction result (dam side), 114b Prediction result (power plant side), 120 measurement control information acquisition management unit, 125 learning data generation unit, 130 model learning unit, 135 prediction processing unit, 140 prediction result presentation unit, 145 dust remover control unit, 200 monitoring control device, 210 storage unit, 212 control instruction information, 220 measurement value acquisition and transfer unit, 225 control instruction receiving unit, 230 gate opening / closing unit control unit, 235 first dust remover control unit, 240 second dust remover control unit, S800a prediction model (dam side) learning process, S800b prediction model (power plant side) learning process, S900a prediction process (dam side), S900b prediction process (power plant side), 1000a prediction result presentation screen (dam side), 1000b prediction result presentation screen (power plant side)

Claims

1. An information processing system that is configured using an information processing device having a processor and a storage device, and that manages water intake equipment provided at a water intake port that takes in water to be used for hydroelectric power generation from a dam and that has a first screen for removing debris, A first prediction model is generated, which is a machine learning model trained using time series data of the water level of the dam, the water level of a waterway that guides water taken from the intake to a hydroelectric power plant, and the opening degree of a gate that controls the amount of water taken from the dam that flows into the waterway as a first explanatory variable, and a probability of overflow from the dam as a first objective variable; determining and outputting the first objective variable by inputting the newly acquired first explanatory variable into the first prediction model; Water intake facility management system.

2. An information processing system configured using an information processing device having a processor and a storage device, and configured to manage a water intake facility equipped with a first screen for removing debris, the water intake facility being provided at a water intake port for taking in water for hydroelectric power generation from a dam, comprising: a first prediction model is generated, which is a machine learning model that is trained using time series data of the water level of the dam, the water level of a waterway that guides water taken from the intake to a hydroelectric power plant, and the opening degree of a gate that controls the amount of water taken from the dam that flows into the waterway as a first explanatory variable, and information indicating the probability of overflow from the dam occurring and the predicted time when overflow from the dam will begin as a first objective variable; determining and outputting the first objective variable by inputting the newly acquired first explanatory variable into the first prediction model; Water intake facility management system.

3. An information processing system configured using an information processing device having a processor and a storage device, and configured to manage a water intake facility equipped with a first screen for removing debris, the water intake facility being provided at a water intake port for taking in water for hydroelectric power generation from a dam, comprising: The water intake facility includes a first dust remover that automatically removes dust adhering to the first screen, a first prediction model is generated, which is a machine learning model that is trained using time series data of the water level of the dam, the water level of a waterway that guides water taken from the intake to a hydroelectric power plant, and the opening degree of a gate that controls the amount of water taken from the dam that flows into the waterway as first explanatory variables, and information indicating the probability of overflow from the dam and the need to start the first dust collector as first objective variables; determining and outputting the first objective variable by inputting the newly acquired first explanatory variable into the first prediction model; Water intake facility management system.

4. The water intake facility management system according to claim 3, A communication device is connected to the first dust remover, When the information output as the first objective variable indicates a need to start up the first dust remover, a control instruction for starting up the first dust remover is transmitted to the first dust remover. Water intake facility management system.

5. The water intake facility management system according to claim 1, The water intake facility further includes a second screen for removing debris, the second screen being provided in a water tank that stores the water guided by the waterway and supplies the water to a hydraulic pipeline of the hydroelectric power plant; A second prediction model is generated, which is a machine learning model trained using time series data of the water level of the water tank and the power output of the hydroelectric power plant as a second explanatory variable and a probability of overflow from the water tank as a second objective variable; The newly acquired second explanatory variable is input to the second prediction model to determine and output the second objective variable. Water intake facility management system.

6. A water intake equipment management system as claimed in claim 1, The water intake facility further includes a second screen for removing debris, the second screen being provided in a water tank that stores the water guided by the waterway and supplies the water to a hydraulic pipeline of the hydroelectric power plant; A second prediction model is generated, which is a machine learning model trained using time series data of the water level of the water tank and the power output of the hydroelectric power plant as second explanatory variables, and information indicating the probability of overflow from the water tank and the predicted time when overflow from the water tank will start as second objective variables; The newly acquired second explanatory variable is input to the second prediction model to determine and output the second objective variable. Water intake facility management system.

7. A water intake equipment management system as claimed in claim 1, The water intake facility further includes a second screen for removing dust and dirt, the second screen being provided in a water tank that stores the water guided by the waterway and supplies the water to a hydraulic pipeline of the hydroelectric power plant, and a second dust remover that automatically removes dust and dirt adhering to the second screen; A second prediction model is generated, which is a machine learning model trained using time series data of the water level of the water tank and the power output of the hydroelectric power plant as second explanatory variables, and information indicating the probability of overflow from the water tank and the need to start the second dust collector as second objective variables; The newly acquired second explanatory variable is input to the second prediction model to determine and output the second objective variable. Water intake facility management system.

8. The water intake facility management system according to claim 7, A communication device is connected to the second dust remover, When the information output as the second objective variable indicates a need to start the second dust remover, a control instruction for starting the second dust remover is transmitted to the second dust remover. Water intake facility management system.

9. A method for managing water intake facilities provided at an intake port for taking in water for hydroelectric power generation from a dam and having a first screen for removing debris, comprising the steps of: An information processing device having a processor and a storage device, A step of generating a first prediction model, which is a machine learning model trained using time series data of the water level of the dam, the water level of a waterway that guides water taken from the intake to a hydroelectric power plant, and the opening degree of a gate that controls the amount of water taken from the dam that flows into the waterway as a first explanatory variable, and the probability of overflow from the dam as a first objective variable; a step of inputting the newly acquired first explanatory variable into the first prediction model to obtain and output the first objective variable; A water intake facility management method that implements the above.

10. The water intake facility management method according to claim 9, The water intake facility further includes a second screen for removing debris, the second screen being provided in a water tank that stores the water guided by the waterway and supplies the water to a hydraulic pipeline of the hydroelectric power plant; The information processing device, A step of generating a second prediction model, which is a machine learning model trained using time series data of the water level of the water tank and the power output of the hydroelectric power plant as second explanatory variables and a probability of overflow from the water tank as a second objective variable; a step of inputting the newly acquired second explanatory variable into the second prediction model to obtain and output the second objective variable; A water intake facility management method that implements the above.

Images