Power generation amount prediction device, power generation amount prediction system, and program

Abstract

The power generation amount prediction device according to an embodiment of the present invention is used to predict a power generation amount of a power generation facility in a given future period, and includes: a past power generation amount data acquisition unit that acquires past power generation amount data that is actual performance data of a past power generation amount of the power generation facility; a weather information acquisition unit that acquires weather information predicted for the given period; a facility information acquisition unit that acquires facility information including at least position information indicating a setting position of the power generation facility; and a power generation amount prediction unit that predicts a power generation amount based on a machine learning model input with the past power generation amount data, the weather information, and a power generation amount predicted based on a physical model that predicts the power generation amount based on the weather information and the facility information.

Description

Technical Field

[0001] This invention relates to a power generation prediction device, a power generation prediction system, and a program. Background Technology

[0002] In renewable energy-based power generation, power generation fluctuates according to weather conditions, necessitating power generation forecasting to stabilize the power system. Power generation operators formulate power generation plans for their owned power plants. When there is a discrepancy between the planned and actual power generation, costs are incurred for the difference. Therefore, high-accuracy power generation forecasting is crucial to improving the profitability of power generation operators.

[0003] Patent document 1 discloses a power generation prediction device that uses a model built through machine learning to calculate the predicted power generation value. The machine learning uses explanatory variables that include weather forecast information from multiple grids, including the prediction location, and target variables that correspond to the power generation of natural energy.

[0004] <Prior art documents> <Patent Documents> Patent Document 1: International Publication No. 2020 / 203854 Summary of the Invention <Problem to be solved by this invention> Even with sophisticated model construction in the estimation model of power generation, there are limits to the improvement of the prediction accuracy of power generation due to factors such as the setting conditions of power generation equipment.

[0005] The purpose of the power generation prediction device involved in this invention is to provide a power generation prediction device capable of performing high-precision power generation prediction in order to solve the aforementioned problems.

[0006] <Methods for solving problems> An embodiment of the present invention relates to a power generation prediction apparatus that predicts the power generation of a power generation device during a given future period. The apparatus comprises: a past power generation data acquisition unit that acquires past power generation data of the power generation device; a meteorological information acquisition unit that acquires meteorological information anticipated during the given period; an equipment information acquisition unit that acquires at least equipment information including location information indicating the installation location of the power generation device; and a power generation prediction unit that predicts power generation based on a machine learning model. This machine learning model is input to the past power generation data, the meteorological information, and a power generation predicted based on a physical model, wherein the physical model predicts the power generation based on the meteorological information and the equipment information.

[0007] <The Effects of the Invention> According to the present invention, high-precision power generation prediction is possible. Attached Figure Description

[0008] Figure 1 This is a block diagram illustrating an example of the overall structure of the power generation prediction system according to the first embodiment of the present invention.

[0009] Figure 2 It is a diagram used to illustrate the sun's altitude and azimuth.

[0010] Figure 3 This is a block diagram illustrating an example of the hardware structure of the power generation prediction device according to the first embodiment of the present invention.

[0011] Figure 4 This is a block diagram illustrating an example of the functional structure of the power generation prediction device according to the first embodiment of the present invention.

[0012] Figure 5 This is a flowchart illustrating the power generation prediction process in the power generation prediction device according to the first embodiment of the present invention.

[0013] Figure 6 It is used for explanation Figure 5 The diagram shows the details of the processing steps S104 to S106.

[0014] Figure 7A It is a graph that plots the actual power generation value on the vertical axis and the power generation value on the horizontal axis based on the date and time, along with the predicted value from the machine learning model and the predicted value from the physical model.

[0015] Figure 7B This is a graph showing the MAE% value of the predicted power generation value relative to the actual power generation value.

[0016] Figure 8A This is a graph showing the effect of using solar position information in the machine learning model of the power generation prediction device according to the first embodiment of the present invention, and the reduction in power generation caused by the shadow of buildings.

[0017] Figure 8B This is a graph showing the effect of using solar position information in the machine learning model of the power generation prediction device according to the first embodiment of the present invention, and the reduction in power generation caused by the presence of mountain shadows.

[0018] Figure 9 This is a block diagram illustrating an example of the overall structure of the power generation prediction system according to the second embodiment of the present invention.

[0019] Figure 10 This is a diagram showing an example of a power curve in a wind power generation device.

[0020] Figure 11 This is a block diagram illustrating an example of the functional structure of the power generation prediction device according to the second embodiment of the present invention.

[0021] Figure 12 This is a flowchart illustrating the power generation prediction process in the power generation prediction device according to the second embodiment of the present invention.

[0022] Figure 13 It is used for explanation Figure 12 The diagram shows the details of the processing steps S204 to S206.

[0023] Figure 14 This is a diagram showing the result of power generation prediction in the power generation prediction device according to the second embodiment of the present invention.

[0024] Figure 15 This is a configuration diagram of the energy supply system using the power generation prediction system involved in this embodiment. Detailed Implementation

[0025] Hereinafter, embodiments of the present invention will be described with reference to the accompanying drawings. Furthermore, in this specification and the accompanying drawings, constituent elements having substantially the same functional structure are labeled with the same reference numerals, and repeated descriptions are omitted.

[0026] [First Implementation] <Overall Structure> Regarding the overall structure of the power generation prediction system 1 in the first embodiment, refer to... Figure 1 The explanation will be provided later. Figure 1 This is a block diagram illustrating an example of the overall structure of the power generation prediction system 1 according to the first embodiment of the present invention. The power generation prediction system 1 according to this embodiment is used as a power generation prediction system 1 that utilizes renewable energy sources such as solar power generation, whose power generation is affected by weather or natural conditions.

[0027] The estimation method for the power generation of a solar power generation system using engineering methods based on physical models is difficult to eliminate factors such as the influence of building shadows caused by the installation conditions of the solar power generation equipment. Furthermore, there is a risk of deviations arising from the spatial representativeness of the weather forecast data used as input to the estimation model, the weather conditions at the installation location of the solar power generation equipment, or the degradation of the solar panels. Therefore, the power generation prediction system 1 according to this embodiment has been discovered, capable of solving such problems. Hereinafter, the power generation prediction system 1 will be described as an example of a system used in the case of solar power generation.

[0028] As shown in the figure, the power generation prediction system 1 includes a power generation prediction device 10 and a terminal device 20. The power generation prediction device 10 and the terminal device 20 are connected via a communication network N1, such as a LAN (Local Area Network) or the Internet, in a manner that enables data communication.

[0029] The power generation prediction system 1 is connected to an external data source 30 via a communication network N1 in a manner that enables data communication. The external data source 30 includes a power generation management system 31, a meteorological information system 32, an equipment information management system 33, and a solar position calculation system 34.

[0030] External data source 30 is not limited to these and may include other data sources. For example, external data source 30 may also include other data sources that accumulate data obtained from the power generation management system 31, the meteorological information system 32, the equipment information management system 33, and the solar position calculation system 34, and process that data to provide the data.

[0031] The power generation prediction device 10 is an information processing device such as a personal computer, workstation, or server that predicts the power generation of power generation equipment in a given future period. The power generation prediction device 10 obtains the data required for prediction from an external data source 30 and generates a prediction model for power generation. The power generation prediction device 10 sends the predicted power generation results to a terminal device 20.

[0032] Terminal device 20 is an information processing terminal such as a personal computer, smartphone, or tablet terminal operated by the user of the power generation prediction system 1. Terminal device 20 receives the power generation prediction results from the power generation prediction device 10 and displays the prediction results to the user.

[0033] The power generation management system 31 sends past power generation data, which serves as historical performance data of the power generation equipment, to the power generation prediction device 10. The given period can be, for example, the past 10 years, but is not limited to this; various periods can be specified. The power generation management system 31 is capable of accumulating data related to power generation at multiple locations.

[0034] The meteorological information system 32 is an information processing system that provides meteorological information. Meteorological information may include, for example, forecast data such as air pressure, wind speed, temperature, relative humidity, precipitation, cloud cover, and sunshine duration. The meteorological information system 32 sends meteorological forecast data for the predicted date and time of future power generation for a given period to the power generation forecasting device 10. The given period may be, for example, 3 days or 1 week, but is not limited to these; it can be a variety of periods.

[0035] The equipment information management system 33 sends information related to the power generation equipment to the power generation prediction device 10. In the case of solar power generation equipment, information related to the solar panel includes the solar panel's elevation angle θ1 and azimuth angle φ1, power generation capacity, power regulator capacity, conversion efficiency, and temperature coefficient. Additionally, the information related to the solar panel may also include location information such as the latitude and longitude of the equipment where the solar panel is located.

[0036] The solar position calculation system 34 is an information processing system for calculating solar position information. For example, the solar position calculation system 34 calculates the sun's position based on the latitude and longitude of the location where the solar power generation equipment is installed, as well as date and time information, thereby calculating the solar position information. Specifically, the solar position calculation system 34 calculates the sun's altitude angle θ2 and azimuth angle φ2 relative to the solar panel. The solar position calculation system 34 can be a device for calculating solar position information or a program.

[0037] use Figure 2 This section explains the solar altitude angle θ2 and azimuth angle φ2 included in the solar position information. Figure 2 This diagram illustrates the solar altitude angle θ2 and azimuth angle φ2. The altitude angle θ2 is referenced to the horizon and can range from -90° to 90°. Similarly, the azimuth angle φ2 is referenced to true north and can range from 0° to 360°.

[0038] also, Figure 1 The overall structure of the power generation prediction system 1 shown is an example, and various system configurations are possible depending on the application or purpose. For example, one or more of the power generation prediction device 10 and the terminal device 20 may be included in the power generation prediction system 1. For example, the power generation prediction device 10 may also be implemented using multiple computers, or it may be implemented as a cloud computing service. For example, the power generation prediction system 1 may also be implemented using a stand-alone computer. Figure 1 The division of devices such as the power generation prediction device 10 and the terminal device 20 shown is an example.

[0039] <Hardware Structure> Regarding the hardware structure of the power generation prediction device 10 in the first embodiment, refer to... Figure 3 The explanation will be provided later. Figure 3 This is a block diagram illustrating an example of the hardware structure of the power generation prediction device 10 according to the first embodiment of the present invention.

[0040] like Figure 3As shown, the power generation prediction device 10 includes a CPU (Central Processing Unit) 41, a ROM (Read Only Memory) 42, a RAM (Random Access Memory) 43, an HDD (Hard Disk Drive) 44, an input device 45, a display device 46, a communication I / F (Interface) 47, and an external I / F 48. The CPU 41, ROM 42, and RAM 43 form a so-called computer. The various hardware components of the power generation prediction device 10 are interconnected via a bus 49. Additionally, the input device 45 and the display device 46 can also be connected to an external I / F 48.

[0041] CPU 41 is a computing device that reads programs or data from storage devices such as ROM 42 or HDD 44 into RAM 43 and performs processing, thereby realizing the overall control or function of the power generation prediction device 10. The power generation prediction device 10 may also have a GPU (Graphics Processing Unit) in addition to or in place of CPU 41.

[0042] ROM42 is an example of a non-volatile semiconductor memory (storage device) that can retain programs or data even when the power is disconnected. ROM42 functions as the main storage device for storing various programs and data required by CPU41 to execute various programs installed in HDD44. Specifically, ROM42 stores boot programs such as BIOS (Basic Input / Output System) and EFI (Extensible Firmware Interface) that are executed when the power generation prediction device 10 is started, as well as OS (Operating System) settings, network settings, and other data.

[0043] RAM43 is an example of a volatile semiconductor memory (storage device) whose program or data is erased when power is disconnected. RAM43 is, for example, DRAM (Dynamic Random Access Memory) or SRAM (Static Random Access Memory). RAM43 provides a working area that expands when various programs installed in HDD44 are executed by CPU41.

[0044] HDD44 is an example of a non-volatile storage device that stores programs or data. The programs or data stored in HDD44 include an operating system (OS) that forms the basic software for controlling the power generation prediction device 10 as a whole, and application programs that provide various functions on the OS. Alternatively, the power generation prediction device 10 can replace HDD44 by using a storage device that uses flash memory as the storage medium (e.g., SSD: Solid State Drive, etc.).

[0045] Input device 45 is a touch panel, operation key or button, keyboard or mouse, microphone, etc., used for users to input various signals.

[0046] The display device 46 consists of a liquid crystal or organic EL (Electro-Luminescence) display for displaying images, and a speaker for outputting sound data such as sound.

[0047] Communication I / F47 is an interface connected to the communication network for data communication of the power generation prediction device 10.

[0048] The external I / F48 is an interface to external devices. External devices include driver devices 50, etc.

[0049] The drive device 50 is a device for setting the recording medium 51. The recording medium 51 includes media that record information optically, electrically, or magnetically, such as CD-ROMs, floppy disks, and optical discs. Alternatively, the recording medium 51 may also include semiconductor memories that record information electrically, such as ROMs and flash memory. Therefore, the power generation prediction device 10 can read from and / or write to the recording medium 51 via an external I / F 48.

[0050] Alternatively, various programs installed on the HDD44 can be installed by placing the distributed recording medium 51 into a drive device 50 connected to an external I / F 48, and having the drive device 50 read the various programs recorded on the recording medium 51. Alternatively, various programs installed on the HDD44 can also be installed via communication I / F 47 by downloading them from a network different from the communication network.

[0051] <Functional Structure> Figure 4This is a block diagram illustrating an example of the functional structure of the power generation prediction device 10 according to the first embodiment of the present invention. As shown, the power generation prediction device 10 includes a past power generation data acquisition unit 11, a weather information acquisition unit 12, an equipment information acquisition unit 13, a solar position information acquisition unit 14, and a power generation prediction unit 15. The power generation prediction device 10 predicts the power generation of power generation equipment during a given future period. The given period may be, for example, 3 days or 1 week, but is not limited to these and can be of various types.

[0052] The past power generation data acquisition unit 11 acquires past power generation data from the power generation management system 31, which serves as historical data on past power generation from the power generation equipment. Past power generation data may, for example, be data related to power generation during a given period based on solar panels.

[0053] The meteorological information acquisition unit 12 acquires the expected meteorological information for a given period from the meteorological information system 32. The meteorological information may include air pressure, wind speed, air temperature, relative humidity, precipitation, cloud cover, and sunshine duration.

[0054] The equipment information acquisition unit 13 acquires equipment information from the equipment information management system 33, which includes at least location information indicating the installation location of the power generation equipment. The equipment information may also be equipment information related to the solar panel, including at least one of the following: latitude and longitude of the solar panel installation location, angle of the solar panel, power generation capacity of the solar panel, capacity of the power regulator, conversion efficiency of the power conversion device, and temperature coefficient of the solar panel.

[0055] The solar position information acquisition unit 14 acquires solar position information, which indicates the position of the sun during a given period, based on the installation position of the power generation equipment. The solar position information includes the solar altitude angle θ2 and azimuth angle φ2, based on the installation position of the aforementioned solar panel.

[0056] The power generation prediction unit 15 can predict power generation based on a machine learning model that incorporates past power generation data and meteorological information. Furthermore, the power generation prediction unit 15 can also predict power generation based on a machine learning model that incorporates solar position information, past power generation data, and meteorological information. Moreover, the power generation prediction unit 15 can also predict power generation based on a physical model that incorporates solar position information, meteorological information, and equipment information.

[0057] The power generation prediction unit 15 can also determine the power generation for a given period based on the power generation predicted by the physical model, even without past power generation data from the power generation equipment. When predicting power generation based on the physical model, the power generation prediction unit 15 can make predictions based on the decomposition of solar radiation, the calculation of incident light, the temperature estimation of the cells contained in the solar panel, the estimation of DC power generation, and the estimation of AC power generation.

[0058] The power generation prediction unit 15 can perform preprocessing on the acquired data. For example, the power generation prediction unit 15 can also interpolate the data if power generation data for a certain moment is missing. Regarding data interpolation, for example, the most recent measurement value for that moment can be used to interpolate the missing value.

[0059] The power generation prediction unit 15 can use the preprocessed data to generate a machine learning model for predicting power generation. At this time, the power generation prediction unit 15 learns from multiple teacher data based on a predetermined prediction algorithm, thereby generating machine learning models corresponding to each combination of teacher data and prediction algorithm.

[0060] Teacher data can use meteorological data such as air pressure, wind speed, temperature, relative humidity, precipitation, cloud cover, and sunshine duration input into the machine learning model, as well as solar position information such as the sun's altitude angle θ2 and azimuth angle φ2 relative to the solar panel, and the power generation prediction value from the physical model described later, as positive solution data, and the actual power generation value used. That is, the meteorological information input into the machine learning model includes at least one of air pressure, wind speed, temperature, relative humidity, precipitation, cloud cover, and sunshine duration.

[0061] Furthermore, the power generation prediction unit 15 can also generate multiple machine learning models with different explanatory variables based on the combination of teacher data and prediction algorithms. Regarding explanatory variables, the available explanatory variables can also be predetermined based on the prediction algorithm.

[0062] Machine learning models are models used to predict electricity generation using statistical methods based on target and explanatory variables. A machine learning model includes a prediction algorithm that at least predicts electricity generation. For example, a machine learning model could be a random forest, LightGBM (Light Gradient Boosting Machine), or XGBoost (eXtreme Gradient Boosting).

[0063] The power generation prediction unit 15 predicts power generation using a physical model based on meteorological and equipment information, and calculates the physical model prediction value. The physical model prediction value can be, for example, a power generation prediction every 10 minutes, every 30 minutes, etc., but is not limited to this. Information related to the solar panel includes information such as the solar panel's elevation angle, azimuth angle, power regulator capacity, conversion efficiency, solar panel temperature coefficient, and location information such as the latitude and longitude of the solar panel's location. Meteorological information input into the physical model includes at least one of wind speed, air temperature, and solar radiation.

[0064] The power generation prediction unit 15 is capable of decomposing solar irradiance, calculating incident light, estimating the temperature of the cells contained in the solar panel, estimating DC power generation, and estimating AC power generation. The solar irradiance decomposition breaks down global horizontal irradiance (GHI) into diffuse horizontal irradiance (DHI) and direct normal irradiance (DNI). The incident light calculation takes into account reflection at the solar panel surface and wavelength dependence of atmospheric absorption.

[0065] The temperature estimation of the cells contained in a solar panel is primarily based on the sun's position, solar irradiance, air temperature, and wind speed. The estimation of DC power generation is performed by converting the results of solar irradiance decomposition, incident light calculation, and cell temperature estimation into DC power generation within the solar panel. The estimation of AC power generation is performed by converting the estimated DC power generation into AC power generation using an inverter model.

[0066] The physical model prediction value generated by the power generation prediction unit 15 may not use data related to the power generation of a given period based on the solar panels. In this case, power generation prediction can be performed even in situations where there is no data related to past power generation, such as at locations where new solar panels are installed.

[0067] The power generation prediction unit 15 uses a generated machine learning model and a physical model to predict power generation. Therefore, the power generation prediction unit 15 can perform power generation predictions that exclude factors such as the influence of building shadows caused by the installation conditions of the solar panels. Furthermore, the power generation prediction unit 15 can perform power generation predictions that exclude deviations caused by the spatial representativeness of meteorological information, the weather conditions at the solar panel installation location, and deviations caused by solar panel degradation. Additionally, the power generation prediction unit 15 can perform predictions every 30 minutes, for example, but is not limited to this.

[0068] The power generation prediction unit 15 can also perform certain processing to make the predicted power generation into output data and output it.

[0069] <Processing Procedure> Here, we will explain the inferences made based on the availability of data related to past power generation. Figure 5 This is a flowchart illustrating the power generation prediction process in the power generation prediction device 10 according to the first embodiment of the present invention.

[0070] First, the power generation prediction device 10 acquires weather information for a given future period, equipment information including the location information of the power generation equipment, and solar position information, which are the targets of the prediction (step S101). Next, the power generation prediction unit 15 predicts the power generation based on a physical model input with the acquired information (step S102).

[0071] Here, if a machine learning model exists regarding the power generation equipment being predicted (No in step S103), the power generation prediction unit 15 obtains a power generation prediction value based on the physical model (step S104). Then, the power generation prediction unit 15 performs a power generation prediction based on the machine learning model (step S105). Furthermore, the power generation prediction unit 15 outputs the predicted power generation data (step S106).

[0072] In the absence of a machine learning model for the solar panels being predicted ("Yes" in step S103), the power generation prediction unit 15 outputs power generation data predicted based on the physical model (step S107). Additionally, cases where no machine learning model exists include locations where new power generation equipment has been installed and where past power generation data is unavailable.

[0073] Figure 6 It is used for explanation Figure 5 The diagram illustrates the processing details of steps S104 to S106. As shown, past power generation data, meteorological information, solar position information, and a physical model are input into the machine learning model. The physical model receives meteorological information and equipment information. Meteorological information input into the machine learning model includes, for example, air pressure, wind speed, air temperature, relative humidity, precipitation, cloud cover, and sunshine duration. Meteorological information input into the physical model includes, for example, wind speed, air temperature, and sunshine duration. Furthermore, equipment information input into the physical model includes, for example, the latitude and longitude of the solar panel's location, the angle of the solar panel, the capacity of the power conversion device included in the solar power generation equipment, the conversion efficiency of the power conversion device, and the temperature coefficient.

[0074] Machine learning-based power generation prediction requires past power generation data. Furthermore, machine learning models, by using information about the sun's position for prediction, can account for the sun's location, eliminating the influence of shadows from mountains and buildings, thus improving prediction accuracy. Physical models, on the other hand, do not require past power generation data, allowing predictions even without it, based on factors such as the installation of new solar power equipment.

[0075] <Prediction Accuracy> The prediction accuracy of the power generation prediction device 10 in this embodiment will be explained with reference to Figures 7 and 8. Furthermore, in the following text, the MAE% value is used as an evaluation index for prediction error. The MAE% value is the average absolute error standardized from the maximum power generation (kWh) of the solar power generation equipment. The maximum power generation is the larger of the power generation capacity of the solar panel and the capacity of the power regulator.

[0076] The MAE% value is calculated according to the following formula (1). In formula (1), i is the 30-minute time slot number, n is the predicted time slot number, and E... obs It refers to the actual power generation performance, E pred It is a forecast of electricity generation, E max It represents the maximum power generation capacity of a solar power generation system. Furthermore, it can be said that the smaller the MAE% value, the higher the prediction accuracy.

[0077]

Formula 1

[0078] like Figure 7B As shown, regarding the MAE% value of the predicted power generation relative to the actual power generation, the physical model's prediction was 6.1% over the entire period, while the machine learning model's prediction was 4.9%. Furthermore, in any given month, the machine learning model's prediction was smaller than the physical model's prediction. This demonstrates that the machine learning model's prediction is closer to the actual value than the physical model's prediction.

[0079] Figure 8A , Figure 8B This is a graph showing the effect of using solar position information in the machine learning model of the power generation prediction device 10 according to the first embodiment of the present invention. The vertical axis represents power generation and the horizontal axis represents date and time, showing the actual power generation value, the physical model prediction value, the prediction value of the machine learning model without using solar position information, and the prediction value of the machine learning model using solar position information. Figure 8A This illustrates the reduction in power generation caused by the presence of building shadows. Figure 8B The diagram illustrates the decrease in power generation caused by the presence of mountain shadows, with the vertical axis representing power generation and the horizontal axis representing the passage of time.

[0080] Figure 8A The data shown represents two days of data, with the majority of power generation occurring during the daytime. As illustrated, it can be seen that power generation predictions based on machine learning models are closer to actual values ​​than those based on physical models.

[0081] Figure 8A The MAE% values ​​of the predicted power generation relative to the actual power generation under the following conditions are as follows: the prediction value of the physical model is 7.3%, the prediction value of the machine learning model without using solar position information is 5.1%, and the prediction value of the machine learning model using solar position information is 4.8%. This shows that the accuracy of power generation prediction is improved by using the machine learning model with solar position information.

[0082] Figure 8B This is data for half a day. Power generation increases as noon approaches. In the morning, as indicated by the arrows in the graph, power generation is low due to the influence of mountain shadows. As shown in the figure, it can be seen that power generation predictions based on machine learning models are closer to actual values ​​than those based on physics models. In addition, such as... Figure 8B As shown, even when using a machine learning model, the power generation prediction using solar position information is closer to the actual value than the prediction without solar position information.

[0083] <Effects of the Implementation Method> The power generation prediction device 10 of this embodiment predicts power generation based on a machine learning model that incorporates past power generation data, meteorological information, and solar position information. Furthermore, since the power generation prediction device 10 predicts power generation based on a machine learning model that incorporates predictions from a physical model, it can eliminate prediction biases caused by environmental factors such as meteorological data and deterioration of power generation equipment. Therefore, the power generation prediction device 10 of this embodiment enables highly accurate power generation prediction.

[0084] [Second Implementation] <Overall Structure> Regarding the overall structure of the power generation prediction system 1 in the second embodiment, refer to... Figure 9 The explanation will be provided later. Figure 9 This is a block diagram illustrating an example of the overall structure of the power generation prediction system 1 according to the second embodiment of the present invention. The power generation prediction system 1 according to this embodiment will be described as an example of a system used for wind power generation. In addition, components that are the same as those described in the first embodiment will be given the same reference numerals, and repeated descriptions will be omitted.

[0085] The power generation prediction system 1 is connected to an external data source 30 via a communication network N1 in a manner capable of data communication. The external data source 30 includes a power generation management system 31, a meteorological information system 32, and an equipment information management system 33. Unlike the first embodiment, the power generation prediction system 1 involved in this embodiment does not include a solar position calculation system 34.

[0086] The equipment information management system 33 sends information related to the power generation equipment to the power generation prediction device 10. In the case of wind power generation equipment, the information related to the power generation equipment can include, for example... Figure 10 The information shown, including the power curve of the wind turbine, depicts the relationship between wind speed and power output, i.e., wind turbine equipment information. Power output can be either the amount of electricity generated or the output of the generated power.

[0087] <Functional Structure> Figure 11 This is a block diagram illustrating an example of the functional structure of the power generation prediction device 10 according to the second embodiment of the present invention. As shown, the power generation prediction device 10 includes a past power generation data acquisition unit 11, a weather information acquisition unit 12, an equipment information acquisition unit 13, and a power generation prediction unit 15.

[0088] The equipment information acquisition unit 13 acquires information from the equipment information management system 33, including at least the relationship between wind speed and power output shown in the power curve of the wind power generation equipment, i.e., wind turbine equipment information.

[0089] The power generation prediction unit 15 can predict power generation based on a machine learning model that incorporates past power generation data and meteorological information. Furthermore, the power generation prediction unit 15 can also predict power generation based on a machine learning model that incorporates wind turbine equipment information, past power generation data, and meteorological information.

[0090] The power generation prediction unit 15 can also perform preprocessing on the acquired data. For example, the power generation prediction unit 15 can also interpolate the data if power generation data for a certain moment is missing. The data interpolation can be performed by using the most recent measurement value for that moment to interpolate the missing value.

[0091] The power generation prediction unit 15 performs power generation prediction based on meteorological information and equipment information using a physical model, and calculates the physical model prediction value. In the physical model prediction method according to this embodiment, wind speed is input, and the output is based on, for example... Figure 9 The power generation is estimated from a power curve as shown. The physical model prediction can be, for example, a power generation prediction every 10 minutes, every 30 minutes, etc., but is not limited to this.

[0092] <Processing Procedure> Here, we will explain the inferences made based on the availability of data related to past power generation. Figure 12 This is a flowchart illustrating the power generation prediction process in the power generation prediction device 10 according to the second embodiment of the present invention.

[0093] First, the power generation prediction device 10 acquires weather information for a given future period, including equipment information such as the location information of the power generation equipment, which is the target of the prediction (step S201). Next, the power generation prediction unit 15 predicts the power generation based on a physical model input with the acquired information (step S202).

[0094] Here, if a machine learning model exists regarding the power generation equipment being predicted (No in step S203), the power generation prediction unit 15 obtains a power generation prediction value based on the physical model (step S204). Then, the power generation prediction unit 15 performs a power generation prediction based on the machine learning model (step S205). Furthermore, the power generation prediction unit 15 outputs the predicted power generation data (step S206).

[0095] In the absence of a machine learning model for information about the wind turbine equipment being predicted ("Yes" in step S203), the power generation prediction unit 15 outputs power generation data predicted based on the physical model (step S207). Additionally, cases where no machine learning model exists include locations where new power generation equipment has been installed and where there is no past power generation data.

[0096] Figure 13 It is used for Figure 12The diagram below explains the processing details of steps S204 to S206. As shown in the diagram, past power generation data, meteorological information, and a physical model are input into the machine learning model. The physical model is input with meteorological information and equipment information. The meteorological information input into the machine learning model includes, for example, air pressure, wind speed, air temperature, relative humidity, precipitation, cloud cover, and sunshine duration. The meteorological information input into the physical model includes, for example, wind speed. In addition, the equipment information input into the physical model includes, for example, wind turbine equipment information, and may also include power curves.

[0097] Machine learning-based power generation prediction requires past power generation data. However, physical models, which do not require past power generation data, can predict power generation even without it, based on factors such as the installation of new wind power equipment.

[0098] <Prediction Accuracy> Regarding the prediction accuracy of the power generation prediction device 10 in this embodiment, refer to... Figure 14 The following explanation will be provided. Furthermore, as in the explanation of the first embodiment, the MAE% value will be used as the evaluation metric for prediction error in the following text.

[0099] Figure 14 This is a diagram showing the results of power generation prediction in the power generation prediction device 10 according to the second embodiment of the present invention. Figure 14 The graph, with the vertical axis representing power generation and the horizontal axis representing date and time, shows the actual power generation value, the predicted value based on the machine learning model, and the predicted value based on the physical model. The graph shows that the predicted value based on the machine learning model is closer to the actual value than the predicted value based on the physical model.

[0100] [Other Implementation Methods] The power generation prediction device 10 according to this embodiment can be used not only for solar power generation and wind power generation, but also for power generation projects based on renewable energy sources such as hydropower, where power generation is affected by weather or natural conditions. Furthermore, the power generation prediction device 10 can be used in systems that control energy resources such as power generation equipment utilizing various natural energy sources, large-scale storage batteries capable of supplying stored electricity, and hydrogen power generation, and adjust the balance between power supply and demand. In addition, there is an embodiment of a power generation prediction method executed by the power generation prediction device 10 according to this embodiment. Specific examples will be used to describe this in the following description.

[0101] Figure 15 This is a configuration diagram of an energy supply system 100 that uses the power generation prediction system 1 described in this embodiment. Figure 15The energy supply system 100 shown includes a power operator system 2 comprising a power generation forecasting system 1, a demand-side information processing terminal 3, and demand-side energy resources 4. The power operator system 2 includes the power generation forecasting system 1, a virtual power plant system 60, a supply and demand management system 61, a transmission and distribution system 62, a centralized power source 63, a distributed power source 64, a battery 65, and a hydrogen power generation device 66. The supply and demand management system 61, transmission and distribution system 62, centralized power source 63, distributed power source 64, battery 65, and hydrogen power generation device 66 are, for example, owned by the power operator of the power operator system 2. Figure 15 The energy resource 4 of the demand side shown includes a solar power generation system 71, a wind power generation system 72, a battery system 73, an electric vehicle 74, a V2H (Vehicle to Home) device 75, and a hydrogen production unit 76. Additionally, the power operator may own part or all of the energy resource 4 of the demand side.

[0102] The virtual power plant system 60 effectively utilizes information and communication technologies such as IoT (Internet of Things) to remotely control the charging and discharging of the energy resource 4 of the demand side via communication networks such as the Internet. The virtual power plant system 60 controls the aggregation of surplus electricity from the energy resource 4 of the demand side, which is connected to the power operator system 2, by remotely controlling the energy resource 4, centralized power supply 63, distributed power supply 64, battery 65, and hydrogen power generation equipment 66. Surplus electricity refers to the electricity generated or stored by the energy resource 4 of the demand side that exceeds the electricity consumed by the demand side.

[0103] The virtual power plant system 60 determines the destination of surplus electricity aggregated from energy resources 4, centralized power sources 63, distributed power sources 64, batteries 65, and hydrogen power generation equipment 66 on the demand side. The virtual power plant system 60 can control the supply of surplus electricity to other demand sides, and has the ability to adjust the supply and demand balance.

[0104] The virtual power plant system 60 can return the revenue obtained from supplying surplus electricity to the selected supply destination to the demand side. For example, the virtual power plant system 60 can also use the revenue obtained from supplying surplus electricity to the selected supply destination as a source of income for purposes such as reducing electricity prices, awarding credits, or providing new services.

[0105] The demand-side information processing terminal 3 is a computer operated by the demand-side, such as a PC (Personal Computer), HEMS (Home Energy Management System), or smartphone. The demand-side information processing terminal 3 displays information received from the virtual power station system 60.

[0106] The supply and demand management system 61 manages the supply and demand of electricity to various demanders based on the power generation forecast by the power generation forecasting system 1, in order to achieve overall optimization of power supply and demand. Furthermore, the transmission and distribution system 62 is, for example, a wholesale electricity market, a capacity market, or a supply and demand adjustment market. A wholesale electricity market is a trading market for buying and selling electricity. A capacity market is a trading market for buying and selling future supply capacity. A supply and demand adjustment market is a trading market for buying and selling adjustment capacity.

[0107] Centralized power source 63 is a large-scale power generation device located far from the location of electricity demand, such as a thermal power plant, hydroelectric power plant, or nuclear power plant. Distributed power source 64 is a smaller-scale power generation device located near the location of electricity demand, such as a solar power plant or a wind power plant. Battery 65 is an energy storage device that stores energy for later use or as an energy regulator. Hydrogen power generation device 66, for example, uses hydrogen supplied from a hydrogen cartridge as an energy source to generate electricity and supplies the generated electricity to the device. When surplus electricity from renewable energy sources can be effectively utilized, the hydrogen generated from the surplus electricity can be stored in the hydrogen power generation device 66. In the hydrogen power generation device 66, hydrogen produced and stored by the hydrogen production device 76 using surplus electricity is used to generate electricity.

[0108] Centralized power source 63, distributed power source 64, battery 65, and hydrogen power generation equipment 66 are examples of multiple energy resources. Furthermore, the virtual power plant system 60 controls the power supply and demand of an energy supply platform that integrates multiple energy resources. Connected to the energy supply platform controlled by the virtual power plant system 60 are examples of energy resources 4 acting as demanders: a solar power generation system 71, a wind power generation system 72, a battery system 73, an electric vehicle 74, a V2H device 75, and a hydrogen production unit 76.

[0109] The solar power system 71 and wind power system 72, as examples of energy resources 4 on the demand side, are examples of renewable energy power generation equipment installed at the demand side. The battery system 73 is an example of an electricity storage device installed in a home or similar location. The V2H device 75 is a system that uses the on-board battery of an electric vehicle 74 as a power source for home use.

[0110] in addition, Figure 15 The energy supply system 100 is one example; it goes without saying that there are various system configurations depending on the purpose or intended use. For example, the virtual power plant system 60 can adopt various structures, such as a structure in which multiple systems work together.

[0111] [Replenish] The functions of the embodiments described above can be implemented by one or more processing circuits included in the power generation prediction device 10. Here, "processing circuit" in this specification includes processors such as CPUs (Central Processing Units) or GPUs (Graphics Processing Units) that are programmed to perform the functions by software, ASICs (Application Specific Integrated Circuits), DSPs (Digital Signal Processors), FPGAs (Field Programmable Gate Arrays), conventional circuit modules, and other devices designed to perform the functions described above.

[0112] <Summary of Implementation Methods> This specification discloses at least the following appendix regarding the power generation prediction device, power generation prediction system, power generation prediction method, and procedure.

[0113] (Note 1) A power generation prediction device for predicting the power generation of power generation equipment in a future given period, comprising: The past power generation data acquisition unit acquires past power generation data of the power generation equipment. The meteorological information acquisition department acquires the meteorological information anticipated during the given period; The equipment information acquisition unit acquires at least equipment information including location information indicating the installation location of the power generation equipment; and The power generation forecasting unit predicts power generation based on a machine learning model. This machine learning model is input into the past power generation data, the meteorological information, and the power generation predicted based on a physical model. The physical model predicts the power generation based on the meteorological information and the equipment information.

[0114] (Note 2) According to the power generation prediction device described in Appendix 1, wherein... The power generation equipment is a solar power generation equipment that includes solar panels. It also includes a solar position information acquisition unit, which acquires solar position information indicating the position of the sun during a given period, based on the installation location of the solar power generation device. The solar position information is also input into the machine learning model.

[0115] (Note 3) According to the power generation prediction device described in Appendix 2, wherein, The solar position information includes the solar altitude angle and azimuth angle based on the installation location of the solar power generation device.

[0116] (Note 4) According to the power generation prediction device described in Appendix 2 or 3, wherein... The device information includes device information related to the solar panel. The equipment information associated with the solar panel includes at least one of the following: the latitude and longitude of the location of the solar panel, the angle of the solar panel, the capacity of the power regulator included in the solar power generation equipment, the conversion efficiency of the power regulator, and the temperature coefficient.

[0117] (Note 5) According to any one of Appendices 1 to 4, the power generation prediction device, wherein... The meteorological information input into the machine learning model includes at least one of air pressure, wind speed, air temperature, relative humidity, precipitation, cloud cover, and sunshine duration.

[0118] (Note 6) According to any one of Appendices 1 to 5, the power generation prediction device, wherein, The meteorological information input into the physical model includes at least one of wind speed, temperature, and sunshine duration.

[0119] (Note 7) According to any one of Appendices 1 to 6, the power generation prediction device, wherein... In the absence of past power generation data from the power generation equipment, the power generation prediction unit determines the power generation predicted based on the physical model as the power generation for the given period.

[0120] (Note 8) According to the power generation prediction device described in Appendix 7, wherein... The power generation prediction unit predicts power generation based on the physical model, using the decomposition of solar radiation, calculation of incident light, estimation of the temperature of the cells contained in the solar panel, estimation of DC power generation, and estimation of AC power generation.

[0121] (Note 9) According to the power generation prediction device described in Appendix 1, wherein... The power generation equipment is a wind power generation equipment that includes a wind turbine. The equipment information acquisition unit acquires at least the information of the wind turbine equipment that represents the relationship between wind power and power generation output, i.e., wind turbine equipment information.

[0122] (Postscript 10) According to the power generation prediction device described in Appendix 9, wherein... The meteorological information input into the physical model includes at least wind speed.

[0123] (Postscript 11) A power generation forecasting system is used to predict the power generation of power generation equipment in a future given period, and it comprises: The past power generation data acquisition unit acquires the past power generation data of the power generation equipment. A meteorological information acquisition unit acquires meteorological information anticipated during the given period; The equipment information acquisition unit acquires at least equipment information including location information indicating the installation location of the power generation equipment; and The power generation prediction unit predicts power generation based on a machine learning model. This machine learning model is inputted with past power generation data, meteorological information, and power generation predictions based on a physical model. The physical model predicts the power generation based on the meteorological information and the equipment information.

[0124] (Postscript 12) A power generation forecasting method, executed by a power generation forecasting device that forecasts the power generation of power generation equipment in a given future period, includes the following steps: The step of obtaining past power generation data of the power generation equipment, i.e., past power generation data; The steps to obtain the anticipated meteorological information for the given period; At least the step of obtaining device information including location information indicating the installation location of the power generation equipment; and The step of predicting power generation based on the input past power generation data, the meteorological information, and a machine learning model that predicts power generation based on a physical model. The physical model predicts the power generation based on the meteorological information and the equipment information.

[0125] (Postscript 13) A non-transitory computer-readable medium storing a program for causing a computer to perform the following processes: Processing past power generation data of power generation equipment; Processing of weather information anticipated for a given future period; At least the processing of acquiring device information including location information representing the installation location of the power generation equipment; and The process of predicting power generation is based on the input of the past power generation data, the meteorological information, and a machine learning model that predicts power generation based on a physical model that predicts power generation according to the meteorological information and the equipment information.

[0126] The embodiments of the present invention have been described in detail above. The present invention is not limited to these embodiments. Various modifications or alterations can be made within the scope of the spirit of the present invention as set forth in the claims.

[0127] This application claims priority to Japanese Patent Application No. 2024-54987, filed with the Japanese Patent Office on March 28, 2024, the entire contents of which are incorporated herein by reference.

[0128] Symbol Explanation 1: Power Generation Forecasting System 10: Power generation prediction device 11: Past power generation data acquisition department 12: Meteorological Information Acquisition Department 13: Equipment Information Acquisition Department 14: Solar Position Information Acquisition Department 15: Power Generation Forecasting Department.

Claims

1. A power generation forecasting device, which forecasts the power generation of power generation equipment during a given future period, comprising: The past power generation data acquisition unit acquires past power generation data of the power generation equipment. The meteorological information acquisition department acquires the meteorological information anticipated during the given period; The equipment information acquisition unit acquires at least equipment information including location information indicating the installation location of the power generation equipment; and The power generation forecasting unit predicts power generation based on a machine learning model. This machine learning model is input into the past power generation data, the meteorological information, and the power generation predicted based on a physical model. The physical model predicts the power generation based on the meteorological information and the equipment information.

2. The power generation prediction device according to claim 1, wherein, The power generation equipment is a solar power generation equipment that includes solar panels. The power generation prediction device also includes a solar position information acquisition unit, which acquires solar position information representing the position of the sun during the given period, based on the installation location of the solar power generation equipment. The solar position information is also input into the machine learning model.

3. The power generation prediction device according to claim 2, wherein, The solar position information includes the solar altitude angle and azimuth angle based on the installation location of the solar power generation device.

4. The power generation prediction device according to claim 2, wherein, The device information includes device information related to the solar panel. The equipment information associated with the solar panel includes at least one of the following: the latitude and longitude of the location of the solar panel, the angle of the solar panel, the capacity of the power regulator included in the solar power generation equipment, the conversion efficiency of the power regulator, and the temperature coefficient.

5. The power generation prediction device according to claim 1, wherein, The meteorological information input into the machine learning model includes at least one of air pressure, wind speed, air temperature, relative humidity, precipitation, cloud cover, and sunshine duration.

6. The power generation prediction device according to claim 1, wherein, The meteorological information input into the physical model includes at least one of wind speed, temperature, and sunshine duration.

7. The power generation prediction device according to claim 1, wherein, In the absence of past power generation data for the power generation equipment, the power generation prediction unit will determine the power generation predicted based on the physical model as the power generation for the given period.

8. The power generation prediction device according to claim 7, wherein, The power generation prediction unit predicts power generation based on the physical model, using the decomposition of solar radiation, calculation of incident light, estimation of the temperature of the cells contained in the solar panel, estimation of DC power generation, and estimation of AC power generation.

9. The power generation prediction device according to claim 1, wherein, The power generation equipment is a wind power generation equipment that includes a wind turbine. The equipment information acquisition unit acquires at least the information of the wind turbine equipment that represents the relationship between wind power and power generation output, i.e., wind turbine equipment information.

10. The power generation prediction device according to claim 9, wherein, The meteorological information input into the physical model includes at least wind speed.

11. A power generation forecasting system, which predicts the power generation of power generation equipment during a given future period, comprising: The past power generation data acquisition unit acquires the past power generation data of the power generation equipment. A meteorological information acquisition unit acquires meteorological information anticipated during the given period; The equipment information acquisition unit acquires at least equipment information including location information indicating the installation location of the power generation equipment; and The power generation prediction unit predicts power generation based on a machine learning model. This machine learning model is input into the past power generation data, the meteorological information, and the power generation predicted based on a physical model. The physical model predicts the power generation based on the meteorological information and the equipment information.

12. A non-transitory computer-readable medium storing a program for causing a computer to perform the following processes: Processing past power generation data of power generation equipment; Processing of weather information anticipated for a given future period; At least the processing of acquiring device information including location information representing the installation location of the power generation equipment; and The process of predicting power generation is based on the input of the past power generation data, the meteorological information, and a machine learning model that predicts power generation based on a physical model that predicts power generation according to the meteorological information and the equipment information.

Images