Prediction system, prediction method, and program

Abstract

To provide a prediction system for predicting the amount of power generation of hydroelectric power generation facilities.SOLUTION: A prediction system includes: a prediction model that includes the amount of power generation (t) in an explanation variable and sets the amount of power generation (t+1) as an object variable when the amount of power generation generated by hydroelectric power generation facilities during a predetermined period (t) is set to the amount of power generation (t) and when a period that follows the period (t) and has the same length as the period (t) is set to the period (t+1), and the amount of power generation during the period (t+1) is set to the amount of power generation (t+1); a prediction unit for successively executing prediction of the amount of power generation for each period during a predetermined prediction object period by using the prediction model and a prediction result by the prediction model; and an output unit for outputting a prediction result of the amount of power generation by the prediction unit.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The present disclosure relates to a system, a method, and a program for predicting the amount of power generated by a hydroelectric power generation facility. [Background technology]

[0002] Electric power companies may enter into power purchase contracts with other power plants that have hydroelectric power generation facilities, and supply electricity from these power plants to the power grid. In a power grid, it is important to maintain a balance between supply and demand in order to suppress fluctuations in power frequency. In order to achieve a stable supply of electricity, the amount of electricity that can be supplied (sold) from the hydroelectric power generation facility to the electric power company is agreed upon in advance between the two parties, and penalties are imposed on the hydroelectric power generation facility if it is unable to supply the planned amount of electricity.

[0003] Patent Document 1 discloses a prediction system for predicting the amount of power generation from hydroelectric power generation facilities, which predicts the amount of overflow power lost due to overflow from water taken for power generation using a prediction model, and predicts the power supply capacity in emergencies, etc. by adding the predicted amount of overflow power to the planned amount of power generation based on a power demand forecast. Patent Document 2 discloses a demand forecasting device that, with regard to electricity demand forecasting, uses the results of forecasting electricity demand at other times where the demand is correlated with the forecast time as part of the input data for demand forecasting at the forecast time, and forecasts electricity demand at the forecast time. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] JP 2015-146065 A [Patent Document 2] Patent No. 5618501 Summary of the Invention [Problem to be solved by the invention]

[0005] Patent documents 1 and 2 disclose technologies for predicting electricity demand and technologies for predicting the amount of electricity generated by hydroelectric power generation facilities based on electricity demand, but do not disclose technologies for accurately predicting the amount of electricity that a hydroelectric power generation facility can supply, regardless of electricity demand.

[0006] The present disclosure provides a prediction system, a prediction method, and a program that can solve the above problems. [Means for solving the problem]

[0007] The prediction system of the present disclosure comprises: When the amount of power generated by a hydroelectric power generation facility in a specified period (t) is defined as power generation amount (t), the period of the same length following the period (t) is defined as period (t+1), and the amount of power generation in the period (t+1) is defined as power generation amount (t+1), the following formula is used: the values ​​of parameters that affect the amount of power generation for each period from the past to the prediction target date to which the period (t+1) belongs, the total value of the values ​​of the parameters for each period from the past to the day before the prediction target date, the amount of power generation for each period from the past to the period (t), and the sum of the amounts of power generation for each period from the past to the period (t). The process of setting the explanatory variables and the target variable as explanatory variables for predicting the power generation amount (t+1) and setting the power generation amount (t+1) as a target variable is repeated while changing the prediction target date, thereby creating a plurality of data combining the explanatory variables and the target variable, and a prediction model created by learning learning data including the data, and a prediction model for a predetermined prediction target period divided into the periods, in which the first period is set as a period (s), the predicted value or actual value of the power generation amount in the period (s) is set as the power generation amount (s), and the predicted value or actual value of the parameter is set as a parameter (s ), the sum of the values ​​of the parameters is the sum of the parameters (s), the sum of the amounts of power generation is the total amount of power generation (s), a predicted value or actual value of the amount of power generation in a period (s+1) following the period (s) is the amount of power generation (s+1), a predicted value or actual value of the parameters is the parameter (s+1), the sum of the values ​​of the parameters is the sum of the parameters (s+1), the sum of the amounts of power generation is the total amount of power generation (s+1), and the amount of power generation in a period (s+2) following the period (s+1) is the amount of power generation (s+2). The system includes a prediction unit that sequentially executes the processes of predicting the power generation amount (s+1) based on the power generation amount (s), the parameter (s), the sum value (s) of the parameter, and the total power generation amount (s), and predicting the power generation amount (s+2) based on the prediction model, the predicted power generation amount (s+1), the parameter (s+1), the sum value (s+1) of the parameter, and the total power generation amount (s+1), and predicts the power generation amount for each of the periods in the prediction target period; and an output unit that outputs the power generation amount prediction result by the prediction unit.

[0008] In the prediction method of the present disclosure, When the amount of power generated by a hydroelectric power generation facility in a specified period (t) is defined as power generation amount (t), the period of the same length following the period (t) is defined as period (t+1), and the amount of power generation in the period (t+1) is defined as power generation amount (t+1), the following equation is obtained: the values ​​of parameters that affect the amount of power generation for each period from the past to the prediction target date to which the period (t+1) belongs, the total value of the values ​​of the parameters for each period from the past to the day before the prediction target date, the amount of power generation for each period from the past to the period (t), and the amount of power generation for each period from the past to the period (t). The sum of the above is set as an explanatory variable for predicting the power generation amount (t+1), and the power generation amount (t+1) is set as a response variable. This process is repeated while changing the period (t+1), thereby creating a plurality of data sets combining the explanatory variables and the response variable. Using a prediction model created by learning learning data including the above data, a predetermined prediction target period is divided into the periods, and the first period is set as the period (s), and the predicted value or actual value of the power generation amount in the period (s) is set as the power generation amount (s), and the predicted value or actual value of the parameter is set as the power generation amount (s). the predicted or actual value of the power generation amount in a period (s+1) following the period (s) is the power generation amount (s+1); the predicted or actual value of the parameter is the parameter (s+1); the sum of the values ​​of the parameter is the sum of the parameters (s+1); the sum of the power generation amount is the total power generation amount (s+1); and the power generation amount in a period (s+2) following the period (s+1) is the power generation amount (s+2). The process of predicting the power generation amount (s+1) based on the prediction model, the power generation amount (s), the parameter (s), the sum of the parameters (s), and the total power generation amount (s), and predicting the power generation amount (s+2) based on the prediction model, the predicted power generation amount (s+1), the parameter (s+1), the sum of the parameters (s+1), and the total power generation amount (s+1) is sequentially performed to predict the power generation amount for each period in the prediction target period, and the prediction result of the power generation amount based on the prediction is output.

[0009] The program of the present disclosure is provided to a computer. When the amount of power generated by a hydroelectric power generation facility in a specified period (t) is defined as power generation amount (t), the period of the same length following the period (t) is defined as period (t+1), and the amount of power generation in the period (t+1) is defined as power generation amount (t+1), the following parameters affecting the amount of power generation for each period from the past to the prediction target date to which the period (t+1) belongs, the total value of the values ​​of the parameters for each period from the past to the day before the prediction target date, the amount of power generation for each period from the past to the period (t), and the amount of power generation for each period from the past to the period (t). and the sum of the amounts of power generated (t+1) are set as explanatory variables for predicting the amount of power generated (t+1), and the process of setting the amount of power generated (t+1) as an objective variable is repeated while changing the prediction target date, thereby creating a plurality of data sets combining the explanatory variables and the objective variable, and using a prediction model created by learning learning data including the data, a predetermined prediction target period is divided into the periods, the first period is set as period (s), the predicted value or actual value of the amount of power generated in the period (s) is set as the amount of power generated (s), and the predicted value or actual value of the parameters is set as the predicted value or actual value of the parameters. the value of the parameter is a parameter (s), the sum of the values ​​of the parameters is the sum of the parameters (s), the sum of the amounts of power generation is the total amount of power generation (s), a predicted value or actual value of the amount of power generation in a period (s+1) following the period (s) is the amount of power generation (s+1), the predicted value or actual value of the parameter is a parameter (s+1), the sum of the values ​​of the parameters is the sum of the parameters (s+1), the sum of the amounts of power generation is the total amount of power generation (s+1), and the amount of power generation in a period (s+2) following the period (s+1) is the amount of power generation (s+2). a process of predicting the power generation amount (s+1) based on the prediction model, the power generation amount (s), the parameter (s), the sum of the parameters (s), and the total power generation amount (s), and predicting the power generation amount (s+2) based on the prediction model, the predicted power generation amount (s+1), the parameter (s+1), the sum of the parameters (s+1), and the total power generation amount (s+1), predicting the power generation amount for each period in the prediction target period, and outputting a prediction result of the power generation amount based on the prediction. Execute the command. Effect of the Invention

[0010] According to the above-described prediction system, prediction method, and program, it is possible to predict the amount of power generated by a hydroelectric power generation facility. [Brief description of the drawings]

[0011] [Figure 1] FIG. 1 is a functional block diagram illustrating a configuration example of a prediction system according to an embodiment. [Diagram 2] FIG. 2 is a diagram showing an overview of a prediction model for power generation amount according to the embodiment. [Diagram 3] FIG. 11 is a diagram illustrating an example of explanatory variables of a prediction model according to the embodiment. [Figure 4] 13 is a flowchart illustrating an example of a process for creating a prediction model according to the embodiment. [Diagram 5] 10 is a flowchart illustrating an example of a process for predicting a power generation amount according to the embodiment. [Figure 6] 1 is a diagram used to explain a process for predicting a power generation amount according to an embodiment; [Figure 7] FIG. 2 is a diagram illustrating an example of a hardware configuration of a prediction system according to an embodiment. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0012] <Embodiment> Hereinafter, the prediction system according to the embodiment will be described in detail with reference to FIGS. (composition) FIG. 1 is a functional block diagram illustrating a configuration example of a prediction system according to an embodiment. The prediction system 10 is a system that predicts the amount of power that a hydroelectric power generation facility can generate. The prediction system 10 predicts the amount of power generation on a target prediction date based on the amount of power generation in a predetermined period immediately prior to the target prediction date. As described below, by using the prediction system 10 to perform sequential predictions, it is possible to predict the amount of power generation on a daily basis up to a distant future, such as one month or six months ahead, or for every hour, every 30 minutes, etc.

[0013] As shown in the figure, the prediction system 10 includes a data acquisition unit 11, a prediction model creation unit 12, a prediction unit 13, a storage unit 14, and an output unit 15. The data acquisition unit 11 acquires parameters necessary for predicting the amount of power generation, data necessary for creating a prediction model for predicting the amount of power generation, and the like. The prediction model creation unit 12 processes data necessary for creating a prediction model to create learning data, and creates a prediction model 141 by machine learning or the like using the learning data. The prediction unit 13 predicts the amount of power generated by the hydroelectric power generation facility for each unit period in the prediction target period. The storage unit 14 stores the parameters and learning data acquired by the data acquisition unit 11, the prediction model 141, and the like. The output unit 15 outputs the predicted power generation amount predicted by the prediction unit 13 to an electronic file or the like.

[0014] Next, the prediction model 141 will be described with reference to FIG. 2 and FIG. FIG. 2 shows an overview of the prediction model 141. The prediction model 141 is a prediction model that is created by learning the relationship between explanatory variables and objective variables, with weather information on the prediction target date (t+1), an operation plan on the prediction target date (t+1), and the power generation amount on the day before the prediction target date (t) as explanatory variables, and the power generation amount by the hydroelectric power generation facility on the prediction target date (t+1) as an objective variable. t in FIG. 2 represents a day. In other words, when the weather information and operation plan on the prediction target date and the power generation amount on the day before the prediction target date are input to the prediction model 141, the prediction model 141 outputs a predicted value of the power generation amount generated by the hydroelectric power generation facility on the prediction target date. The inventor has found that there is almost no fluctuation in the power generation amount between the day before and the day of the prediction in the case of a water intake type hydroelectric power generation facility. In order to ensure the accuracy of the prediction, the inventor has conceived the prediction model 141 as shown in FIG. 2 by adding weather information (t+1) and an operation plan (t+1) to the explanatory variables in addition to the power generation amount on the previous day (t). The inventors also confirmed that the power generation amount (t+2) on the day (t+2) following the prediction target day (t+1) can be accurately predicted by replacing the explanatory variable power generation amount (t) with the predicted power generation amount (t+1) and predicting the power generation amount using the prediction model 141. That is, the prediction model 141 enables sequential prediction, for example, to predict the power generation amount on the day based on the actual power generation amount yesterday, to predict the power generation amount on the next day using the power generation amount prediction on the day, and to predict the power generation amount two days later based on the power generation amount prediction on the next day. By using the prediction system 10 to predict the power generation amount sequentially, it is possible to predict the power generation amount up to the distant future, such as one month ahead or six months ahead.

[0015] Next, the explanatory variables will be described in detail with reference to Fig. 3. As shown in Fig. 3, the explanatory variables of the prediction model 141 can be roughly divided into weather information, an operation plan, and an amount of power generation. The weather information is, for example, rainfall. Since the amount of power generated by hydroelectric power generation is related to the amount of water in the river where water is taken, rainfall is used as an explanatory variable. Since changes in the amount of water in a river occur with a delay after rainfall, the accuracy of prediction is improved by using the amount of rainfall or the total amount of rainfall for the past few days from the prediction target date as an explanatory variable. As weather information for explanatory variables, (1) the amount of rainfall for each day from the prediction target date and several days prior, and (2) the total amount of rainfall from the day before the prediction target date to the several days prior to the prediction target date are used. For example, when predicting the amount of power generation on the day, the actual value of the amount of rainfall for the past few days is used. For example, when predicting the amount of power generation one week from now, the predicted value of the amount of rainfall for several days going back one week from now is used. The actual value and the predicted value of the amount of rainfall can be obtained from the Japan Meteorological Agency, a company that provides weather forecast information, etc. For example, the amount of rainfall to be input as an explanatory variable may be calculated based on the actual value and the predicted value of the amount of rainfall observed at multiple observation stations around the location where the hydroelectric power generation facility is located.

[0016] An operation plan is, for example, the operating time of a hydroelectric power generation facility. Since the amount of power generated by hydroelectric power generation is related to the operating time, the operating time is used as an explanatory variable. As operation plan information for the explanatory variables, (1) the operating time for each day from the prediction target day and several days prior, and (2) the total operating time from the day before the prediction target day to several days prior are used. Past operating time can be determined from actual operating data held by the hydroelectric power generation facility, and future operating time can be determined from the operation plan of the hydroelectric power generation facility. By using the total operating time from several days prior, it is possible to make predictions that include trends in the operation plan rather than using only the operating time for the day, thereby improving prediction accuracy.

[0017] The operation plan is information about the calendar in which the hydroelectric power generation facility operates. Because hydroelectric power generation is affected by the seasons, calendar information such as the month and day of the week is used as explanatory variables.

[0018] The power generation amount is the actual value of the power generation amount generated by the hydroelectric power generation facility or the predicted power generation amount value predicted using the prediction model 141. As the power generation amount of the explanatory variables, (1) the power generation amount (actual value or predicted value) for each day from the day before to several days before the prediction target day, and (2) the total power generation amount from the day before to several days before the prediction target day are used. The actual value of the past power generation amount can be grasped from the actual data of the power generation amount held by the hydroelectric power generation facility, and the future power generation amount can be obtained by sequentially predicting it using the prediction model 141.

[0019] In the above description, the unit period for power generation prediction is one day, but the length of the unit period may be one hour or 30 minutes. That is, for example, the prediction model 141 may be created to output a predicted value of power generation for every 30 minutes on the prediction target day (t+1). In that case, for example, the explanatory variables may be as follows: The weather information used is (1) the amount of rainfall for every 30 minutes on the prediction date and for several days prior, and (2) the total amount of rainfall from the day before to the day before the prediction date. The operation plan information used is (1) the operation time for every 30 minutes on the prediction date and for several days prior, (2) the total operation time from the day before to the day before the prediction date, and (3) calendar information. The power generation amount used is (1) the power generation amount (actual value or predicted value) for every 30 minutes on each day from 30 minutes prior to the prediction date and for several days prior, and (2) the total power generation amount from 30 minutes prior to the prediction date and for several days prior. When predicting the power generation amount for the next 30 minutes, the predicted power generation amount value for the previous 30 minutes is used.

[0020] In the above description, the prediction model 141 predicts the amount of power generation on the prediction target day, but the prediction model 141 may be created to output the difference between the amount of power generation on the prediction target day and the amount of power generation on the previous day. That is, the prediction model may be created with the objective variable being the amount of power generation on the prediction target day (t+1) minus the amount of power generation on the previous day (t). In addition, although "several days ago" is described in FIG. 3, for example, the period indicated by the number of days ago in the upper row (rainfall for each day) and the number of days ago in the lower row (total rainfall) of the rainfall column may be different. The same applies to the number of days ago for the operation time and the amount of power generation. The period indicated by the number of days ago described in the upper row of the rainfall, operation time, and amount of power generation may be different from each other. The same applies to the number of days ago described in the lower row.

[0021] (operation) Next, the operation of the prediction system 10 will be described with reference to FIGS. (Predictive model creation process) First, the process of creating a prediction model will be described. FIG. 4 is a flowchart illustrating an example of a process for creating a prediction model according to the embodiment. First, the data acquisition unit 11 acquires data for creating the prediction model 141 (step S1). For example, the data acquisition unit 11 acquires the amount of rainfall, operation time, and power generation amount for each day in a predetermined period in the past. The data acquisition unit 11 records these data in the storage unit 14. Next, the prediction model creation unit 12 performs preprocessing to create learning data (step S2). For example, the prediction model creation unit 12 extracts rainfall data for the prediction target date D1 and for several days before D1 from the data acquired in step S1, with a past day "D1" as the prediction target date, and sets these as explanatory variables for the prediction target date D1. The prediction model creation unit 12 also calculates the total rainfall from the day before the prediction target date D1 to the several days before, and sets the calculated total value as an explanatory variable for the prediction target date D1. The prediction model creation unit 12 also extracts the operation time and power generation amount for the prediction target date D1 and for several days before D1, and further calculates the total operation time and the total power generation amount from the day before the prediction target date D1 to the several days before, and sets these as explanatory variables for the prediction target date D1. The prediction model creation unit 12 also sets calendar information (month and day of the week) for the prediction target date D1 as explanatory variables for the prediction target date D1. Furthermore, the prediction model creation unit 12 extracts the power generation amount for the prediction target date D1 from the data acquired in step S1, and sets this value as the objective variable for the prediction target date D1. The prediction model creation unit 12 creates a large number of combinations of explanatory variables and objective variables, that is, learning data, while changing the prediction target date D1. The prediction model creation unit 12 writes and stores the created large number of learning data in the storage unit 14.

[0022] Next, the prediction model creation unit 12 creates a prediction model 141 (step S3). The prediction model creation unit 12 uses the learning data created in step S12 to learn the relationship between explanatory variables and objective variables by a method such as random forest to create the prediction model 141. The prediction model creation unit 12 writes and stores the prediction model 141 in the storage unit 14. The method of constructing the prediction model 141 is not limited to random forest, and may be regression analysis or other machine learning methods.

[0023] (Power generation forecast processing) Next, the flow of the prediction process will be described. FIG. 5 is a flowchart illustrating an example of a process for predicting the amount of power generation according to the embodiment. First, the data acquiring unit 11 accepts settings of a prediction target period and a unit period (step S11). The prediction target period indicates the period from the beginning to the end of the period for which the amount of power generation is predicted, and the unit period indicates the length of the period for which the amount of power generation is predicted in the prediction target period. For example, a user inputs information to the prediction system 10 instructing to predict the amount of power generation for each day as the unit period from the prediction date (the day when the prediction is performed) to three days later as the prediction target period. The data acquiring unit 11 records the settings of "today to three days later" as the prediction target period and "one day" as the unit period in the storage unit 14.

[0024] Next, the data acquisition unit 11 acquires data necessary for prediction (step S12). Here, FIG. 6 is referred to. It is assumed that the day on which prediction is performed is October 31st, and the "several days before" described with reference to FIG. 3 is always three days before. Then, data necessary for prediction on October 31st, which is the first day of the prediction target period, are the power generation amount, weather information, and actual values ​​of the operation plan from October 28th to October 30th (white circles), and the weather information and predicted values ​​of the operation plan from October 31st to November 2nd (black circles). On the other hand, data necessary for prediction on November 3rd, which is the last day of the prediction target period, are the predicted value of the power generation amount from October 31st to November 2nd (black circles), and the weather information and predicted values ​​of the operation plan from October 31st to November 3rd (black circles). The user inputs the power generation amount, weather information, and actual values ​​of the operation plan from October 28 to October 30, and the weather information and predicted values ​​of the operation plan from October 31 to November 3 to the prediction system 10. The data acquisition unit 11 acquires these data as data necessary for prediction, and records them in the storage unit 14. The predicted value of the power generation amount from October 31 to November 2 is calculated by the following process.

[0025] Next, the prediction unit 13 sets initial values ​​to variables m and t (step S13). Here, m is the difference between the prediction date (=today) and the prediction target date, and t is the prediction date. In this example, since the prediction start date is the prediction date, the prediction unit 13 sets m=0 and t=the prediction date (October 31). Next, the prediction unit 13 predicts the power generation amount Y(t+m) in the first prediction target period of the prediction target period (step S14). In this example, the prediction unit 13 predicts the power generation amount Y (October 31). The prediction unit 13 inputs the power generation amount, weather information, and actual values ​​of the operation plan from October 28 to October 30, and the weather information and predicted values ​​of the operation plan for October 31 to the prediction model 141. The prediction model 141 outputs the power generation amount on October 31. The power generation amount output by the prediction model 141 is the power generation amount Y (October 31). The prediction unit 13 records the power generation amount Y(t+m) in the memory unit 14.

[0026] Next, the prediction unit 13 adds 1 to the variable m (step S15). In this example, m=0+1=1. Next, the prediction unit 13 predicts the power generation amount Y(t+m) using the predicted value of the power generation amount, the power generation amount Y(t+m-1) (step S16). That is, the prediction unit 13 predicts the power generation amount for the next unit period (November 1) using the predicted value of the power generation amount for the previous unit period (October 31). The prediction unit 13 inputs the weather information and the actual value of the operation plan for October 29 to October 31, the weather information and the predicted value of the operation plan for November 1, the actual value of the power generation amount for October 29 to October 30, and the predicted value of the power generation amount for October 31 to the prediction model 141. The prediction model 141 outputs the power generation amount for November 1. The power generation amount output by the prediction model 141 is the power generation amount Y(t+m=November 1). The prediction unit 13 records the power generation amount Y(t+m−1) in the storage unit 14.

[0027] Next, the prediction unit 13 judges whether m is equal to or greater than N (step S17). Here, N is a value obtained by subtracting 1 from the number of unit periods included in the prediction target period. In the present example, N=3. If m is less than N (step S17; No), the prediction unit 13 repeats the process from step S15 onwards. For example, in the present example, if m=1, the prediction unit 13 predicts the amount of power generation on November 2nd using the actual value of the amount of power generation on October 30th and the predicted value of the amount of power generation from October 31st to November 1st. Furthermore, in the next loop, the prediction unit 13 predicts the amount of power generation on November 3rd using the predicted value of the amount of power generation from October 31st to November 2nd. In this way, the prediction unit 13 sequentially repeats the process of predicting the next unit period using the prediction result for the previous unit period.

[0028] When m is equal to or greater than N (step S17; Yes), the output unit 15 outputs the predicted value of the amount of power generated for each unit period in the prediction target period (step S18). For example, the output unit 15 outputs the predicted value of the amount of power generated by the hydroelectric power generation facility for each day from October 31 to November 3 to a display device, an electronic file, or the like.

[0029] In the above explanation, an example of predicting the amount of power generation per day over a period of four days has been given. However, for example, a prediction model 141 created to output predicted power generation values ​​every 30 minutes may be used to predict the amount of power generation per 30 minutes up to three days in advance, or a prediction model of power generation per day up to one month in advance may be used.

[0030] (effect) As described above, the prediction system 10 can accurately predict the amount of power that the hydroelectric power generation facility can supply by using data that the hydroelectric power generation facility can measure (past power generation amount and actual values ​​of operation time and future operation plan), weather information available on the market (actual values ​​and predicted values), and the prediction model 141. In addition, by performing sequential prediction, it is possible to predict the amount of power generation over a long period of time. As a result, when selling the power generated by the hydroelectric power generation facility, it is possible to determine the amount of power generation to be sold based on a highly accurate power generation prediction, and it is possible to reduce the risk of receiving a penalty in the case where the promised amount of power cannot be supplied. In addition, the power generation prediction of this embodiment does not require a power demand prediction, and can predict the amount of power generation according to the capacity of the facility based on the actual results. For example, the inventor constructed the prediction model 141, performed sequential prediction according to the flowchart of FIG. 5, and evaluated the results, and it was confirmed that the amount of power generation can be predicted with an accuracy of about 95% for predictions several days ahead and about 90% for predictions one month ahead.

[0031] FIG. 7 is a diagram illustrating an example of a hardware configuration of the prediction system according to the embodiment. The computer 900 includes a CPU 901 , a main memory device 902 , an auxiliary memory device 903 , an input / output interface 904 , and a communication interface 905 . The prediction system 10 described above is implemented in a computer 900. Each of the above-described functions is stored in the auxiliary storage device 903 in the form of a program. The CPU 901 reads the program from the auxiliary storage device 903, loads it in the main storage device 902, and executes the above-described processing in accordance with the program. The CPU 901 also reserves a storage area in the main storage device 902 in accordance with the program. The CPU 901 also reserves a storage area in the auxiliary storage device 903 for storing data being processed in accordance with the program.

[0032] In addition, a program for realizing all or part of the functions of the prediction system 10 may be recorded on a computer-readable recording medium, and the program recorded on the recording medium may be read into a computer system and executed to perform processing by each functional unit. The "computer system" here includes hardware such as an OS and peripheral devices. In addition, if a WWW system is used, the "computer system" also includes a homepage providing environment (or display environment). In addition, the "computer-readable recording medium" refers to a portable medium such as a CD, DVD, or USB, and a storage device such as a hard disk built into a computer system. In addition, if the program is distributed to the computer 900 through a communication line, the computer 900 that receives the program may expand the program in the main storage device 902 and execute the above processing. In addition, the above program may be for realizing part of the functions described above, and may further be capable of realizing the functions described above in combination with a program already recorded in the computer system. The prediction system 10 may be composed of multiple computers 900.

[0033] As described above, several embodiments according to the present disclosure have been described, but all of these embodiments are presented as examples and are not intended to limit the scope of the invention. These embodiments can be implemented in various other forms, and various omissions, substitutions, and modifications can be made without departing from the gist of the invention. These embodiments and their modifications are included in the scope of the invention and its equivalents as described in the claims, as well as in the scope and gist of the invention.

[0034] <Additional Notes> The prediction system 10, the prediction method, and the program described in the embodiment can be understood, for example, as follows.

[0035] (1) The prediction system 10 according to the first aspect includes a prediction model 141 in which, when the amount of power generated by a hydroelectric power generation facility in a predetermined period (t) is defined as a power generation amount (t), a period of the same length following the period (t) is defined as a period (t+1), and the amount of power generated in the period (t+1) is defined as a power generation amount (t+1), the prediction model 141 includes the amount of power generation (t) as an explanatory variable and the amount of power generation (t+1) as a target variable, and a prediction model 142 in which the first period when a predetermined prediction target period is divided into the periods is defined as a period (s), the amount of power generation in the period (s) is defined as a power generation amount (s), and the amount of power generation in the period (s+1) following the period (s) is defined as a power generation amount (s). The prediction unit 13 sequentially executes the process of predicting the power generation amount (s+1) based on the prediction model and an actual value of the power generation amount (s), and predicting the power generation amount (s+2) based on the prediction model and the predicted power generation amount (s+1), where the power generation amount in period (s+1) is the power generation amount (s+1), and the power generation amount in period (s+2) following period (s+1) is the power generation amount (s+2), and predicts the power generation amount (s+2) based on the prediction model and the predicted power generation amount (s+1). The prediction unit 13 predicts the power generation amount for each period in the prediction target period, and outputs the power generation amount prediction result by the prediction unit. For example, it was found that the amount of power that can be generated by intake-type hydroelectric power generation facilities does not vary greatly between the previous day and the current day. Using a prediction model created based on this knowledge, with the amount of power generated (t) as the explanatory variable and the amount of power generated (t+1) as the objective variable, and the predicted values ​​from this prediction model, it is possible to predict the amount of power generated (s+1) for period (s+1) from the amount of power generated (s) for period (s), predict the amount of power generated (s+2) for period (s+2) from the amount of power generated (s+1), and so on, by successively predicting the amount of power generated for the next period, it is possible to predict the amount of power generated for each unit period up to period (s+N), which is the last period of the prediction target period. This makes it possible to predict the amount of power that can be supplied according to the performance and capacity of the facility, regardless of the power demand, etc.

[0036] (2) A prediction system 10 according to a second aspect is the prediction system 10 of (1), wherein the explanatory variables of the prediction model 141 include, in addition to the power generation amount (t), the amount of rainfall in a specified range around the hydroelectric power generation facility during the period (t) and the operating time of the hydroelectric power generation facility during the period (t) (Figure 2). This makes it possible to improve the accuracy of prediction of the amount of power generation.

[0037] (3) A prediction system 10 according to a third aspect is the prediction system 10 of (1) to (2), and the explanatory variables of the prediction model 141 include, in addition to the power generation amount (t), the power generation amount for each period in a predetermined period before the period (t) including the period (t), a total of the power generation amount for each period before the period (t) including the period (t), the rainfall in a predetermined area around the hydroelectric power generation facility for each period in a predetermined period before the period (t) including the period (t) and the rainfall in the period (t+1), a total of the rainfall for each period before the period (t) including the period (t), the operating time of the hydroelectric power generation facility for each period in a predetermined period before the period (t) including the period (t) and the operating time in the period (t+1), a total of the operating time for each period before the period (t) including the period (t), and information on the month and day of the week of the period (t+1) (Figure 3). This makes it possible to improve the accuracy of prediction of the amount of power generation.

[0038] (4) A prediction system 10 according to a fourth aspect is the prediction system 10 of (1) to (3), further comprising a prediction model creation unit that creates the prediction model by learning the relationship between the explanatory variables and the objective variable. This allows the creation of a prediction model 141.

[0039] (5) A prediction system 10 according to a fifth aspect is a prediction system 10 according to any one of (1) to (4), wherein the objective variable of the prediction model 141 is the difference between the power generation amount (t) and the power generation amount (t+1) instead of the power generation amount (t+1). The prediction model 141 can also be created by setting the response variable to the difference between the power generation amount (t) and the power generation amount (t+1).

[0040] (6) A prediction method according to a sixth aspect includes the steps of: assuming that an amount of power generated by a hydroelectric power generation facility in a predetermined period (t) is a power generation amount (t); assuming that a period of the same length following the period (t) is a period (t+1); and assuming that the amount of power generated in the period (t+1) is a power generation amount (t+1), the prediction method uses a prediction model including the amount of power generation (t) as an explanatory variable and having the amount of power generation (t+1) as a target variable, and divides a predetermined prediction target period into the periods, the first period is defined as a period (s), the amount of power generation in the period (s) is defined as a power generation amount (s), and When the amount of power generation in the subsequent period (s+1) is designated as power generation (s+1) and the amount of power generation in the period (s+2) following the period (s+1) is designated as power generation (s+2), the process of predicting the amount of power generation (s+1) based on the prediction model and the actual value of the amount of power generation (s) and predicting the amount of power generation (s+2) based on the prediction model and the predicted amount of power generation (s+1) is sequentially performed to predict the amount of power generation for each period in the prediction target period and output the prediction result of the amount of power generation based on the prediction.

[0041] (7) A program according to a seventh aspect of the present invention is a program for causing a computer to execute a prediction model including the amount of power generation (t) generated by a hydroelectric power generation facility in a predetermined period (t) as the amount of power generation (t), a period of the same length following the period (t) as the period (t+1), and the amount of power generation in the period (t+1) as the amount of power generation (t+1), the program includes the amount of power generation (t) as an explanatory variable and uses the amount of power generation (t+1) as an objective variable to calculate the amount of power generation (t) in a predetermined prediction period divided into the periods, the first period being a period (s), the amount of power generation in the period (s) being the amount of power generation (s), and the amount of power generation in the period (s) being the amount of power generation (s), and the prediction model includes the amount of power generation (t) as an explanatory variable and uses the amount of power generation (t+1) as an objective variable to calculate the amount of power generation (t) in a predetermined prediction period divided into the periods, the first period being a period (s), the amount of power generation in the period (s) being the amount of power generation (s), and the prediction model includes the amount of power generation (t) in a predetermined prediction period divided into the periods, ... and the prediction model includes the amount of power generation (t) in a predetermined prediction period divided into the periods, and the prediction model includes the amount of power generation (t) When the amount of power generation in a period (s+1) following the period (s+1) is taken as power generation amount (s+1), and the amount of power generation in a period (s+2) following the period (s+1) is taken as power generation amount (s+2), the process of predicting the amount of power generation (s+1) based on the prediction model and an actual value of the amount of power generation (s), and predicting the amount of power generation (s+2) based on the prediction model and the predicted amount of power generation (s+1) is sequentially performed, and a process of predicting the amount of power generation for each period in the prediction target period and outputting a prediction result of the amount of power generation based on the prediction is executed. [Explanation of symbols]

[0042] 10. Prediction System 11 Data acquisition section 12. Prediction model creation section 13. Prediction section 14...Storage section 141...Prediction model 15 Output section 900...Computer 901··CPU 902...Main memory 903...Auxiliary storage device 904 Input / Output Interface 905 Communication Interface

Claims

1. When the amount of electricity generated by a hydroelectric power generation facility in a given period (t) is defined as the amount of electricity generated (t), the period of the same length following the period (t) is defined as period (t+1), and the amount of electricity generated in the period (t+1) is defined as the amount of electricity generated (t+1), Values ​​of parameters that affect the amount of power generation for each period from the past to the prediction target date to which the period (t+1) belongs; and A total value of the parameter values ​​for each of the periods from the past to the day before the prediction target date; and The amount of power generation for each period from the past to the period (t); A total value of the amount of power generation for each period from the past to the period (t); as explanatory variables for predicting the power generation amount (t+1), and setting the power generation amount (t+1) as a response variable. By repeatedly changing the prediction target date, Creating a plurality of sets of data combining the explanatory variables and the objective variables; A prediction model created by learning learning data including the data; and When a predetermined prediction period is divided into the periods, the first period is defined as a period (s), the predicted value or actual value of the power generation amount in the period (s) is defined as the power generation amount (s), the predicted value or actual value of the parameter is defined as a parameter (s), the sum of the values ​​of the parameters is defined as the sum of the parameters (s), the sum of the power generation amounts is defined as the total power generation amount (s), the predicted value or actual value of the power generation amount in a period (s+1) following the period (s) is defined as the power generation amount (s+1), the predicted value or actual value of the parameter is defined as a parameter (s+1), the sum of the values ​​of the parameters is defined as the sum of the parameters (s+1), and the sum of the power generation amounts is defined as the power generation amount (s). a prediction unit that sequentially executes the processes of predicting the power generation amount (s+1) based on the prediction model, the power generation amount (s), the parameter (s), the sum of the parameters (s), and the total power generation amount (s), and predicting the power generation amount (s+2) based on the prediction model, the predicted power generation amount (s+1), the parameter (s+1), the sum of the parameters (s+1), and the total power generation amount (s+1), and predicts the power generation amount (s+2) based on the prediction model, the predicted power generation amount (s+1), the parameter (s+1), the sum of the parameters (s+1), and the total power generation amount (s+1); and an output unit that outputs a result of the prediction of the power generation amount by the prediction unit; A prediction system comprising:

2. The explanatory variables of the prediction model are: The amount of rainfall in a predetermined area around the hydroelectric power generation facility during the period (t); and The operation time of the hydroelectric power generation facility during the period (t); The prediction system of claim 1 .

3. The explanatory variables of the prediction model are: The parameters include a rainfall amount in a predetermined range around the hydroelectric power generation facility and an operating time of the hydroelectric power generation facility, Furthermore, information on the month and day of the week of the prediction target date, The prediction system of claim 1 .

4. The objective variable of the prediction model is the difference between the power generation amount (t) and the power generation amount (t+1) instead of the power generation amount (t+1). A prediction system according to any one of claims 1 to 3.

5. When the amount of electricity generated by a hydroelectric power generation facility in a given period (t) is defined as electricity generation amount (t), the period of the same length following the period (t) is defined as period (t+1), and the amount of electricity generated in the period (t+1) is defined as electricity generation amount (t+1), Values ​​of parameters that affect the amount of power generation for each period from the past to the prediction target date to which the period (t+1) belongs; and A total value of the parameter values ​​for each of the periods from the past to the day before the prediction target date; and The amount of power generation for each period from the past to the period (t); A sum of the amount of power generation for each period from the past to the period (t); as explanatory variables for predicting the power generation amount (t+1), and setting the power generation amount (t+1) as a response variable. By repeatedly performing the above-mentioned process while changing the period (t+1), A plurality of sets of data combining the explanatory variables and the objective variables are generated, and a prediction model is generated by learning training data including the data, When a predetermined prediction period is divided into the periods, the first period is defined as period (s), the predicted value or actual value of the power generation amount in the period (s) is defined as power generation amount (s), the predicted value or actual value of the parameter is defined as parameter (s), the sum of the values ​​of the parameters is defined as the sum of the parameters (s), the sum of the power generation amounts is defined as the total power generation amount (s), the predicted value or actual value of the power generation amount in a period (s+1) following the period (s) is defined as power generation amount (s+1), the predicted value or actual value of the parameter is defined as parameter (s+1), the sum of the values ​​of the parameters is defined as the sum of the parameters (s+1), and the sum of the power generation amounts is defined as a power generation amount (s+1) is a total power generation amount, and the power generation amount in a period (s+2) following the period (s+1) is a power generation amount (s+2), the power generation amount (s+1) is predicted based on the prediction model, the power generation amount (s), the parameter (s), the sum of the parameters (s), and the total power generation amount (s), and the power generation amount (s+2) is predicted based on the prediction model, the predicted power generation amount (s+1), the parameter (s+1), the sum of the parameters (s+1), and the total power generation amount (s+1). The above processes are sequentially executed to predict the power generation amount for each period in the prediction target period; outputting a prediction result of the power generation amount based on the prediction; Forecasting methods.

6. On the computer, When the amount of power generated by a hydroelectric power generation facility in a given period (t) is defined as power generation amount (t), the period of the same length following the period (t) is defined as period (t+1), and the amount of power generated in the period (t+1) is defined as power generation amount (t+1), Values ​​of parameters that affect the amount of power generation for each period from the past to the prediction target date to which the period (t+1) belongs; and A total value of the parameter values ​​for each of the periods from the past to the day before the prediction target date; and The amount of power generation for each period from the past to the period (t); A sum of the amount of power generation for each period from the past to the period (t); as explanatory variables for predicting the power generation amount (t+1), and setting the power generation amount (t+1) as a response variable. By repeatedly changing the prediction target date, Creating a plurality of sets of data combining the explanatory variables and the objective variables; Using a prediction model created by learning learning data including the data, When a predetermined prediction period is divided into the periods, the first period is defined as period (s), the predicted value or actual value of the power generation amount in the period (s) is defined as power generation amount (s), the predicted value or actual value of the parameter is defined as parameter (s), the sum of the values ​​of the parameters is defined as the sum of the parameters (s), the sum of the power generation amounts is defined as the total power generation amount (s), the predicted value or actual value of the power generation amount in a period (s+1) following the period (s) is defined as power generation amount (s+1), the predicted value or actual value of the parameter is defined as parameter (s+1), the sum of the values ​​of the parameters is defined as the sum of the parameters (s+1), and the sum of the power generation amounts is defined as a power generation amount (s+1) is a total power generation amount, and the power generation amount in a period (s+2) following the period (s+1) is a power generation amount (s+2), the power generation amount (s+1) is predicted based on the prediction model, the power generation amount (s), the parameter (s), the sum of the parameters (s), and the total power generation amount (s), and the power generation amount (s+2) is predicted based on the prediction model, the predicted power generation amount (s+1), the parameter (s+1), the sum of the parameters (s+1), and the total power generation amount (s+1). The above processes are sequentially executed to predict the power generation amount for each period in the prediction target period; a process of outputting a prediction result of the power generation amount based on the prediction; A program that executes the following.

Images