Photovoltaic power generation management device, photovoltaic power generation management system, and photovoltaic power generation management method

Abstract

The purpose of the present invention is to provide a photovoltaic power generation management means with which, for a discretionary time range from a short period to a middle and a long period, it is possible to generate a high-accuracy solar radiation amount or photovoltaic power generation amount prediction on a wide area.　To this end, the photovoltaic power generation management device according to the present invention comprises: a model training unit that trains a first solar radiation amount prediction model and a second solar radiation amount prediction model; a prediction information management unit that acquires prediction request information including prediction target region information that defines a prediction target region and prediction time range information that defines a prediction time range; a prediction unit that selects the first solar radiation amount prediction model or the second solar radiation amount prediction model depending on whether or not the prediction time range information included in the prediction request information satisfies a predetermined time range threshold, and generates solar radiation amount prediction information indicating a predicted solar radiation amount for the prediction time range using the selected solar radiation amount prediction model; and a fluctuation management unit that generates and outputs solar radiation variation amount information indicating the amount of variation in solar radiation.

Description

Photovoltaic power generation management device, photovoltaic power generation management system, and photovoltaic power generation management method

[0001] The present disclosure relates to a photovoltaic power generation management device, a photovoltaic power generation management system, and a photovoltaic power generation management method.

[0002] In recent years, as demand for clean energy has increased, the introduction of renewable energy sources, including solar power plants, into the power grid has expanded. The amount of electricity generated by renewable energy sources fluctuates due to various factors, such as the amount of sunlight and wind direction, making the supply to the power grid prone to instability. For this reason, the spread of renewable energy sources has raised concerns about the stability of the power grid.

[0003] There have been proposals for predicting the amount of power generated by renewable energy sources. For example, U.S. Pat. No. 11,545,830 (Patent Document 1) describes a technology for predicting the amount of solar power generation. Patent Document 1 describes a technology in which "a photovoltaic system can include multiple photovoltaic inverters that convert sunlight into electricity. The amount of power generated by each inverter is measured over a period of time. These measurements can be collected together with other data. The collected measurements can be used to generate an artificial neural network that predicts the output of each inverter based on input parameters. These neural networks can be used to predict the total solar power generation of the photovoltaic system."

[0004] U.S. Patent No. 1,545,830

[0005] Patent Document 1 describes a means for generating a machine learning model that predicts solar power generation by training a neural network using measurements collected from multiple photovoltaic inverters that convert sunlight into electricity.

[0006] Although the method described in Patent Literature 1 can generate a solar power generation forecast for a specific solar power plant, the neural network described in Patent Literature 1 is trained based on the output values ​​of the inverters installed in the solar power plant, and therefore the forecast obtained from the neural network is limited to the specific solar power plant. Therefore, to obtain a solar power generation forecast for a wide area including, for example, multiple solar power plants, a new machine learning model must be trained for each solar power plant, taking into account the characteristics of the inverters installed in the solar power plant, which may limit the versatility of the model. Furthermore, the forecast generated by the neural network described in Patent Literature 1 is a short-term forecast, such as 15 minutes, 1 hour, or 24 hours, and therefore it is difficult to consider the impact of periodic features that appear over long periods, such as seasonal changes, on solar power generation. Furthermore, the neural network described in Patent Literature 1 is trained directly based on the inverter output values ​​and does not take into account the relationship between inverter output and solar radiation, which may limit the robustness of the machine learning model.

[0007] Therefore, the present disclosure aims to provide a solar power generation management means capable of generating highly accurate solar radiation or solar power generation forecasts for a wide area for any time range, whether short-term, medium-term, or long-term.

[0008] In order to solve the above problem, a representative solar power generation management means of the present invention is a solar power generation management device including a processor, a memory, and a storage unit that stores past weather information that characterizes the climate of a specified region and weather forecast information for the specified region, wherein the memory includes a model training unit that trains a first solar radiation prediction model that predicts solar radiation for a first time range and a second solar radiation prediction model that predicts solar radiation for a second time range, a prediction information management unit that acquires prediction request information including prediction target area information that defines a specified prediction target region and prediction time range information that defines the time range of prediction, and when the prediction time range information included in the prediction request information satisfies a specified time range threshold, a solar power generation management unit that uses the first solar radiation prediction model to train the past weather information that characterizes the climate of a specified region and weather forecast information for the specified region. The method includes processing instructions to cause the processor to function as a prediction unit that analyzes past weather information for forecasting that characterizes the climate of the target prediction area to generate solar radiation forecast information indicating the amount of solar radiation predicted in the target prediction area for the time range of the prediction, and if the prediction time range information included in the prediction request information does not satisfy a predetermined time range threshold, uses the second solar radiation forecast model to analyze the past weather information for forecasting and weather forecast information for forecasting the target prediction area obtained from the weather forecast information to generate the solar radiation forecast information indicating the amount of solar radiation predicted in the target prediction area for the time range of the prediction, and a fluctuation management unit that generates and outputs solar radiation fluctuation information indicating the amount of fluctuation in the predicted solar radiation based on the solar radiation forecast information.

[0009] According to the present disclosure, it is possible to provide a photovoltaic power generation management means capable of generating highly accurate solar radiation or photovoltaic power generation forecasts for a wide area for any time range, including short-term, medium-term, and long-term. Other problems, configurations, and advantages will become clear from the description of the following embodiments of the present invention.

[0010] FIG. 1 is a diagram illustrating a computer system for implementing an embodiment of the present disclosure. FIG. 2 is a diagram illustrating an example of the configuration of a photovoltaic power generation management system according to an embodiment of the present disclosure. FIG. 3 is a diagram illustrating a flow of a photovoltaic power generation management process according to an embodiment of the present disclosure. FIG. 4 is a diagram illustrating an example of a process for training a first solar radiation prediction model according to an embodiment of the present disclosure. FIG. 5 is a diagram illustrating an example of a process for training a second solar radiation prediction model according to an embodiment of the present disclosure. FIG. 6 is a diagram illustrating an example of an operation of a photovoltaic power generation amount management unit according to an embodiment of the present disclosure. FIG. 7 is a diagram illustrating an example of the flow of a fluctuation amount information generation process according to an embodiment of the present disclosure. FIG. 8 is a diagram illustrating an example of a user interface screen according to an embodiment of the present disclosure.

[0011] Hereinafter, an embodiment of the present invention will be described with reference to the drawings. However, the present invention is not limited to this embodiment. In the description of the drawings, the same parts are designated by the same reference numerals. Furthermore, although terms such as "first," "second," and "third" may be used to describe various elements or components in this disclosure, it will be understood that these elements or components should not be limited by these terms. These terms are used only to distinguish one element or component from another. Therefore, a first element or component discussed below could also be referred to as a second element or component without departing from the teachings of the inventive concept.

[0012] 1, a computer system 100 for implementing embodiments of the present disclosure will be described. The mechanisms and devices of various embodiments disclosed herein may be applied to any suitable computing system. Major components of the computer system 100 include one or more processors 102, memory 104, a terminal interface 112, a storage interface 113, an I / O (input / output) device interface 114, and a network interface 115. These components may be interconnected via a memory bus 106, an I / O bus 108, a bus interface unit 109, and an I / O bus interface unit 110.

[0013] Computer system 100 may include one or more general-purpose programmable central processing units (CPUs) 102A and 102B, collectively referred to as processors 102. In some embodiments, computer system 100 may include multiple processors, while in other embodiments, computer system 100 may be a single CPU system. Each processor 102 executes instructions stored in memory 104 and may include an on-board cache. In some embodiments, computer system 100 may include a graphics processing unit (GPU) in addition to processors 102. The GPU may be used to speed up processing, such as machine learning models used in solar power generation management application 150, described below.

[0014] In some embodiments, memory 104 may include random-access semiconductor memory, storage devices, or storage media (either volatile or non-volatile) for storing data and programs. Memory 104 may store all or part of the programs, modules, and data structures that implement the functions described herein. For example, memory 104 may store solar power management application 150. In some embodiments, solar power management application 150 may include instructions or descriptions that execute on processor 102 the functions described below.

[0015] In some embodiments, solar power management application 150 may be implemented in hardware via semiconductor devices, chips, logic gates, circuits, circuit cards, and / or other physical hardware devices instead of or in addition to a processor-based system. In some embodiments, solar power management application 150 may include data other than instructions or descriptions. In some embodiments, cameras, sensors, or other data input devices (not shown) may be provided to communicate directly with bus interface unit 109, processor 102, or other hardware in computer system 100.

[0016] Computer system 100 may include a bus interface unit 109 that facilitates communication between processor 102, memory 104, display system 124, and I / O bus interface unit 110. I / O bus interface unit 110 may couple to an I / O bus 108 for transferring data to and from various I / O units. I / O bus interface unit 110 may communicate via I / O bus 108 with multiple I / O interface units 112, 113, 114, and 115, also known as I / O processors (IOPs) or I / O adapters (IOAs).

[0017] Display system 124 may include a display controller, a display memory, or both. The display controller may provide video, audio, or both data to display device 126. Computer system 100 may also include one or more sensors or other devices configured to collect data and provide the data to processor 102.

[0018] For example, computer system 100 may include biometric sensors that collect heart rate data, stress level data, etc., environmental sensors that collect humidity data, temperature data, pressure data, etc., and motion sensors that collect acceleration data, movement data, etc. Other types of sensors may also be used. Display system 124 may be connected to a display device 126, such as a standalone display screen, a television, a tablet, or a handheld device.

[0019] The I / O interface unit provides functionality for communicating with various storage or I / O devices. For example, the terminal interface unit 112 may be attached to user I / O devices 116, such as user output devices such as a video display, a television with speakers, and user input devices such as a keyboard, a mouse, a keypad, a touchpad, a trackball, buttons, a light pen, or other pointing device. A user may use a user interface to enter input data or instructions into the user I / O devices 116 and the computer system 100, and receive output data from the computer system 100, by operating the user input devices. The user interface may be displayed on a display, played through speakers, or printed via a printer via the user I / O devices 116, for example.

[0020] Storage interface 113 may accept one or more disk drives or direct access storage devices 117 (typically magnetic disk drive storage devices, but may also be an array of disk drives or other storage devices configured to appear as a single disk drive). In some embodiments, storage device 117 may be implemented as any secondary storage device. The contents of memory 104 may be stored in storage device 117 and retrieved as needed from storage device 117. I / O device interface 114 may provide an interface to other I / O devices, such as printers, fax machines, etc. Network interface 115 may provide a communications path that allows computer system 100 and other devices to communicate with each other. This communications path may be, for example, network 130.

[0021] In some embodiments, computer system 100 may be a device that receives requests from other computer systems (clients) without a direct user interface, such as a multi-user mainframe computer system, a single-user system, or a server computer. In other embodiments, computer system 100 may be a desktop computer, a portable computer, a laptop, a tablet computer, a pocket computer, a telephone, a smartphone, or any other suitable electronic device.

[0022] Next, a photovoltaic power generation management system according to an embodiment of the present disclosure will be described with reference to FIG. 2 .

[0023] Fig. 2 is a diagram illustrating an example of a configuration of a photovoltaic power generation management system 200 according to an embodiment of the present disclosure. The photovoltaic power generation management system 200 according to an embodiment of the present disclosure is a system for generating highly accurate solar radiation or photovoltaic power generation forecasts for a wide area for any short-term, medium-term, or long-term time range, and mainly includes a photovoltaic power generation management device 210 and a user terminal 260 as shown in Fig. 2. The photovoltaic power generation management device 210 and the user terminal 260 may be connected to each other via a communication network 250.

[0024] The solar power generation management device 210 is a device for generating highly accurate solar radiation or solar power generation forecasts for a wide area for any short-term, medium-term, or long-term time range, and as shown in Fig. 2, mainly includes a memory 220, a storage unit 230, a processor 244, and an input / output unit 246. In an embodiment, the solar power generation management device 210 may be implemented by the computer system 100 shown in Fig. 1.

[0025] The memory 220 may be a memory for storing a solar power generation management application 150 for implementing the functionality of a solar power generation management means according to embodiments of the present disclosure. The solar power generation management application 150 may include processing instructions for implementing the functionality of software modules such as a model training unit 222, a forecast information manager 224, a forecasting unit 226, a solar power generation amount manager 228, and a fluctuation manager 229, as shown in FIG.

[0026] The model training unit 222 is a functional unit for training a machine learning model that predicts solar radiation. In an embodiment, the model training unit 222 may train a machine learning model that predicts solar radiation for each of multiple time ranges. In the present disclosure, a "time range" refers to a predetermined period (length of time). As an example, the model training unit 222 may train a machine learning model that predicts solar radiation for a time range of "three months or more" (e.g., a first solar radiation prediction model) or a machine learning model that predicts solar radiation for a time range of "less than three months" (e.g., a second solar radiation prediction model). As described below, the type and training process of each machine learning model differ depending on the time range for which solar radiation prediction is performed. Note that the details of the processing by the model training unit 222 will be described later, and therefore will not be described here.

[0027] The prediction information management unit 224 is a functional unit that receives, via the user terminal 260, prediction request information characterizing the region to be predicted, and then acquires information used for the prediction from the past weather information DB 236 and the weather forecast information DB 238. The prediction request information may include, for example, prediction target area information defining a specific prediction target area, solar power generation facility information characterizing solar power generation facilities to be installed in the prediction target area, and prediction time range information defining the time range of the prediction. Based on the prediction target area information and prediction time range information included in the prediction request information, the prediction information management unit 224 may acquire prediction-use past weather information (and / or prediction-use weather forecast information) characterizing the climate of the prediction target area from the past weather information DB 236 (and / or weather forecast information DB 238). Note that details of the processing by the prediction information management unit 224 will be described later, and therefore will not be described here.

[0028] The prediction unit 226 is a functional unit that selects an insolation prediction model to be used for insolation prediction from among the machine learning models pre-trained by the model training unit 222, based on the prediction time range information included in the prediction request information, and generates insolation prediction information indicating the amount of insolation predicted in the prediction target area for the time range of prediction by analyzing the past weather information for prediction (and / or the weather forecast information for prediction) using the selected insolation prediction model. In an embodiment, the prediction unit 226 may select an insolation prediction model suitable for the time range specified in the prediction time range information included in the prediction request information. As an example, the prediction unit 226 compares the time range specified in the predicted time range information with a predetermined time range threshold (e.g., "3 months") that has been set in advance, and if the predicted time range information satisfies the time range threshold (e.g., if the time range of "4 months" specified in the predicted time range information exceeds the time range threshold of "3 months"), it generates solar radiation prediction information that indicates the amount of solar radiation predicted in the prediction target area by analyzing the past weather information for prediction using a solar radiation prediction model (e.g., a first solar radiation prediction model) that has been trained to make predictions for a time range of three months or more. On the other hand, the time range specified in the prediction time range information is compared with a predetermined time range threshold (e.g., "3 months"). If the prediction time range information does not satisfy the time range threshold (e.g., if the time range specified in the prediction time range information is "2 months" and exceeds the "3 months" time range threshold), the prediction past weather information is analyzed using a solar radiation prediction model trained to make predictions for time ranges less than three months (e.g., a second solar radiation prediction model) to generate solar radiation prediction information indicating the solar radiation predicted for the prediction target area. In this way, by making predictions using a machine learning model trained to predict solar radiation for a desired time range, highly accurate solar radiation prediction information can be obtained for any time range. Note that details of the processing of the prediction unit 226 will be described later, and therefore will not be described here.

[0029] The photovoltaic power generation amount management unit 228 is a functional unit that generates photovoltaic power generation amount prediction information indicating the amount of photovoltaic power generation predicted in the prediction target area based on the solar radiation prediction information generated by the prediction unit 226 and the photovoltaic power generation facility information included in the prediction request information received by the prediction information management unit 224. In an embodiment, the photovoltaic power generation amount management unit 228 may be a regression model trained to convert the predicted amount of solar radiation into the amount of photovoltaic power generation. Note that the details of the processing by the photovoltaic power generation amount management unit 228 will be described later, and therefore will not be described here.

[0030] The fluctuation management unit 229 is a functional unit that generates and outputs solar radiation fluctuation information indicating fluctuations in predicted solar radiation by analyzing the solar radiation forecast information generated by the forecasting unit 226. In an embodiment, the fluctuation management unit 229 may generate and output solar power generation fluctuation information indicating fluctuations in predicted solar power generation based on the solar power generation forecast information generated by the solar power generation amount management unit 228 and a reference forecast value indicating a standard solar power generation amount in the prediction target area. Here, the solar radiation fluctuation amount and the solar power generation amount may be expressed as a percentage. Note that the details of the processing by the fluctuation management unit 229 will be described later, and therefore will not be described here.

[0031] The memory unit 230 is a memory area for storing various information related to an embodiment of the present disclosure, and may include a past weather information DB 236, a weather forecast information DB 238, and a global solar radiation information DB 239, as shown in FIG. 2 .

[0032] The past weather information DB 236 is a database (DB) that stores past weather information characterizing the climate of a plurality of regions. This past weather information may include information such as past cloud cover, precipitation, and global solar radiation measured in a particular region for a predetermined period. In one embodiment, this past weather information may be obtained from a weather information provider such as the Japan Meteorological Agency and stored in the past weather information DB 236. The weather forecast information DB 238 is a database that stores weather forecast information for a plurality of regions, indicating weather forecasts for the respective regions. This weather forecast information may include information such as cloud cover, precipitation, and wind speed forecasted for a particular region for a predetermined period. In one embodiment, this weather forecast information may be obtained from a weather information provider such as the Japan Meteorological Agency and stored in the weather forecast information DB 238. The global solar radiation information DB 239 is a database that stores global solar radiation information for a plurality of regions. The global solar radiation information may include information on global horizontal irradiance (GHI) measured for a predetermined period of time. Global solar radiation here refers to the amount of global solar radiation energy received per unit area on a horizontal surface. The weather forecast information may be measured at power plants included in a specific region and stored in the storage unit 230.

[0033] The processor 244 is a processing unit for executing processing instructions stored by the memory 220 that define the functionality of each functional unit of the solar power generation management application 150 .

[0034] The input / output unit 246 is a functional unit for receiving information (prediction request information) input to the solar power generation management device 210 and outputting information (such as solar radiation fluctuation information and solar power generation fluctuation information) generated by the solar power generation management device 210. In an embodiment, the input / output unit 246 may include, for example, a keyboard, a mouse, and a display that displays a GUI (Graphical User Interface). In an embodiment, the input / output unit 246 may provide the user terminal 260 with a GUI that inputs and outputs various types of information.

[0035] Communications network 250 may include, for example, a local area network (LAN), a wide area network (WAN), a satellite network, a cable network, a WiFi network, or any combination thereof.

[0036] The user terminal 260 is a terminal device that can be used by a user who requests a prediction from the solar power generation management device 210. By using the user terminal 260, the user can check information on fluctuations in solar radiation, information on fluctuations in solar power generation, and the like output from the solar power generation management device 210. As an example, the user terminal 260 may include, but is not limited to, a smartphone, smartwatch, tablet, personal computer, or the like of a user who has subscribed to a solar power generation management service provided by the solar power generation management system 200. Note that, for convenience of explanation, FIG. 2 illustrates an example of a configuration including one user terminal 260, but the number of user terminals 260 is not limited, and a configuration including multiple user terminals 260 is also possible.

[0037] According to the solar power generation management system 200 of the present disclosure described above, it is possible to provide a solar power generation management means capable of generating highly accurate solar radiation or solar power generation forecasts for a wide area for any time range, including short-term, medium-term, and long-term.

[0038] Next, a flow of a photovoltaic power generation management process according to an embodiment of the present disclosure will be described with reference to FIG. 3 .

[0039] FIG. 3 is a diagram illustrating a flow of a solar power generation management process 300 according to an embodiment of the present disclosure. The solar power generation management process 300 illustrated in FIG. 3 is a process for generating highly accurate solar radiation or solar power generation forecasts for a wide area over any short-, medium-, or long-term time range. The process is executed by the functional units of the solar power generation management system 200 illustrated in FIG. 2 . Note that the solar power generation management process 300 is performed on the assumption that a solar radiation forecasting model for forecasting solar radiation and a power generation conversion model for converting solar radiation into solar power generation have been trained in advance. As described above, each solar radiation forecasting model may be trained to forecast solar radiation for a different time range. For ease of explanation, the following description will be given using an example of a first solar radiation forecasting model that forecasts solar radiation for a first time range (e.g., three months or more) and a second solar radiation forecasting model that forecasts solar radiation for a second time range (e.g., less than three months). However, the present disclosure is not limited thereto, and the number of solar radiation forecasting models and the applicable time range are not particularly limited. The details of the training process for the solar radiation prediction model will be described later, and therefore will not be described here.

[0040] First, in step S302, the prediction information management unit 224 acquires prediction target area information that defines a predetermined prediction target area. This prediction target area information may include, for example, geographic coordinates (latitude and longitude) that define the area for which a prediction is requested. In one embodiment, the prediction information management unit 224 may acquire the prediction target area information by receiving a bounding box that defines the prediction target area on a map from the user via the user terminal 260. Note that the prediction target area defined here includes one or more solar power plants (hereinafter referred to as "power plants"). A "solar power plant" refers to a location where solar power generation equipment, such as solar power generation panels, inverters, and batteries, is installed.

[0041] Next, in step S304, the prediction information management unit 224 acquires photovoltaic power generation facility information characterizing the photovoltaic power generation facilities to be installed in the prediction target area. This photovoltaic power generation facility information may include information indicating the location, number, type, specifications, scale, output (megawatts) relative to a predetermined global solar radiation, and the like, of the photovoltaic power generation facilities to be installed in the prediction target area defined in the prediction target area information acquired in step S302. In some embodiments, the prediction information management unit 224 may receive the photovoltaic power generation facility information from a user via the user terminal 260. In some embodiments, the prediction information management unit 224 may analyze map information of the prediction target area defined in step S302 using a predetermined image processing means to identify the photovoltaic power generation facilities to be installed in the prediction target area and estimate the location, number, type, use, scale, maximum output (megawatts), and the like.

[0042] Next, in step S306, the prediction information management unit 224 selects a power plant from the prediction target area specified in step S302 for which fluctuation information (solar radiation fluctuation information or solar power generation fluctuation information) has not yet been generated.

[0043] Next, in step S308, the prediction information management unit 224 acquires prediction time range information that specifies the time range of the prediction. This prediction time range information is information that indicates the time range for which prediction is desired, and may be, for example, "two months" or "four months." In some embodiments, the prediction information management unit 224 may receive the prediction time range information from a user via the user terminal 260. In other embodiments, this prediction time range information may be set for each power plant. This allows the user to set a different prediction time range for each power plant.

[0044] Next, in step S310, the prediction information management unit 224 determines whether the predicted time range indicated in the predicted time range information acquired in step S308 satisfies a predetermined time range threshold. If the predicted time range indicated in the predicted time range information acquired in step S308 satisfies the predetermined time range threshold, the process proceeds to step S312. On the other hand, if the predicted time range indicated in the predicted time range information acquired in step S308 does not satisfy the predetermined time range threshold, the process proceeds to step S316.

[0045] The time range threshold here is information that defines a specific time range and is used to determine an appropriate solar radiation prediction model and information to be used for the solar radiation prediction. In other words, the time range threshold is a period defined to distinguish between "short-term," "medium-term," "long-term," etc. As an example, the time range threshold is used to determine whether to use a first solar radiation prediction model that predicts solar radiation for a first time range (e.g., three months or more) or a second solar radiation prediction model that predicts solar radiation for a second time range (e.g., less than three months). The time range threshold may be set by a user or based on statistical analysis of past prediction simulations. An example will be described in which the time range threshold is "three months or more." If the time range defined in step S308 is "two months," "two months" does not satisfy the time range threshold of "three months or more." Therefore, the process proceeds to step S316, where solar radiation prediction is performed using a second solar radiation prediction model that predicts solar radiation for a time range less than three months. On the other hand, if the time range specified in step S308 is "four months," "four months" satisfies the time range threshold of "three months or more," so the process proceeds to step S312, where a solar radiation prediction is performed using the first solar radiation prediction model that predicts solar radiation for a time range of three months or more.

[0046] If it is determined in step S310 that the prediction time range indicated in the prediction time range information acquired in step S308 satisfies the predetermined time range threshold, in step S312, the prediction information management unit 224 acquires, from the past weather information DB 236, past weather information for prediction, which is past weather information to be used for prediction, based on the prediction time range indicated in the prediction time range information acquired in step S308. As a general rule, when predicting solar radiation levels, it is desirable to use past weather information covering a longer time range (in other words, past weather information that meets a predetermined length criterion) in order to make predictions for a longer time range (e.g., three months or more). Using past weather information covering a longer time range reduces errors and enables more accurate prediction results. Therefore, in this case, the prediction information management unit 224 acquires past weather information for an appropriate time range in accordance with the prediction time range desired by the user.

[0047] More specifically, the forecast information management unit 224 first determines a reference time range, which indicates the time range of past weather information required to keep the error in the solar radiation forecast accuracy within a predetermined tolerance threshold. This tolerance threshold is information that defines an acceptable error and may be set by the user or based on statistical analysis of past forecast simulations. The reference time range indicates the time range of past weather information required to keep the error within the tolerance threshold when forecasting solar radiation for a forecast time range desired by the user. The reference time range may be set based on a predetermined heuristic (e.g., "the reference time range is three times the forecast time range") or based on statistical analysis of past forecast simulations. Next, the forecast information management unit 224 obtains information corresponding to the determined reference time range from the past weather information DB 236 as past weather information for forecasting. This past weather information for forecasting is information indicating past weather (cloud cover, global solar radiation, precipitation, etc.) for the area of ​​the power plant selected in step S306, and is past weather information for forecasting for a time range that meets a specified length criterion (e.g., three months).

[0048] Next, in step S314, the prediction unit 226 analyzes the past weather information for forecasting acquired in step S312 using a first solar radiation forecasting model that predicts solar radiation for a first time range (e.g., three months or more), thereby generating solar radiation forecast information indicating the solar radiation forecasted for the power plant selected in step S306 for the first time range. In this way, highly accurate solar radiation forecast information for a long time range (e.g., three months or more) can be obtained.

[0049] If it is determined in step S310 that the prediction time range indicated in the prediction time range information acquired in step S308 does not satisfy the predetermined time range threshold, in step S316, the prediction information management unit 224 acquires, from the past weather information DB 236, forecast-use past weather information, which is past weather information to be used for prediction, based on the prediction time range indicated in the prediction time range information acquired in step S308, and also acquires forecast-use weather forecast information for the region of the power plant selected in step S306 from the weather forecast information DB 238. As described above, to make predictions for longer time ranges (e.g., three months or more), it is desirable to use past weather information for longer time ranges. On the other hand, to make predictions for shorter time ranges (e.g., less than three months), it is desirable to use weather forecast information provided by, for example, the Japan Meteorological Agency in addition to past weather information. This is because, although the reliability of weather forecasts is limited for long-term periods, the reliability of weather forecasts is sufficiently high for short-term and medium-term periods. Therefore, more accurate prediction results can be obtained by predicting solar radiation based on weather forecast information in addition to past weather information. Therefore, as described in step S312, the forecast information management unit 224 obtains information corresponding to the determined reference time range from the past weather information DB 236 as past weather information for forecasting, and also obtains weather forecast information for forecasting related to the area of ​​the power plant selected in step S306 from the weather forecast information DB 238.

[0050] Next, in step S318, the prediction unit 226 and the prediction information management unit 224 analyze the forecast-use past weather information and forecast-use weather forecast information acquired in step S316 using a second solar radiation prediction model that predicts solar radiation for a second time range (e.g., less than three months), thereby generating solar radiation prediction information that indicates the solar radiation predicted at the power plant selected in step S306 for the second time range. In this way, highly accurate solar radiation prediction information for short- to medium-term time ranges (e.g., less than three months) can be obtained.

[0051] Next, in step S320, the prediction information management unit 224 determines whether or not to convert the solar radiation amount prediction information generated in step S314 or S318 into solar power generation amount information. Whether or not to convert the solar radiation amount prediction information into solar power generation amount information may be determined based on a user instruction input via the user terminal 260, for example. If the solar radiation amount prediction information is to be converted into solar power generation amount information, the process proceeds to step S322. On the other hand, if the solar radiation amount prediction information is not to be converted into solar power generation amount information, the process proceeds to step S324.

[0052] When converting the solar radiation forecast information into solar power generation amount information, in step S322, the solar power generation amount management unit 228 generates solar power generation amount forecast information indicating the amount of solar power generation predicted in the prediction target area, based on the solar radiation forecast information generated in step S314 or S318 and the solar power generation facility information acquired in step S304. In an embodiment, the solar power generation amount management unit 228 may generate the solar power generation amount forecast information by analyzing the solar radiation forecast information using a regression model (power generation conversion model) trained to convert the predicted amount of solar radiation into solar power generation.

[0053] Next, in step S324, the prediction information management unit 224 stores in the storage unit 230 the solar radiation prediction information generated in step S314 or S318, or the solar power generation amount prediction information generated in step S322.

[0054] Next, in step S326, the forecast information management unit 224 checks whether solar radiation forecast information or solar power generation forecast information has been generated for all power plants within the forecast target area defined in the forecast target area information acquired in step S302 and stored in the storage unit 230. If solar radiation forecast information or solar power generation forecast information has been generated for all power plants and stored in the storage unit 230, the process proceeds to step S328. On the other hand, if there are unprocessed power plants, the process returns to step S306, and the processes from step S308 onwards are performed for the unprocessed power plants.

[0055] Next, in step S328, the fluctuation management unit 229 analyzes the solar radiation forecast information generated for each power plant to generate solar radiation fluctuation information indicating fluctuations in the predicted solar radiation. Furthermore, if solar power generation forecast information is available for a specific power plant, the fluctuation management unit 229 analyzes the solar power generation forecast information to generate solar power generation fluctuation information indicating fluctuations in the predicted solar power generation. More specifically, the fluctuation management unit 229 may generate and output solar power generation fluctuation information indicating fluctuations in the predicted solar power generation based on the solar power generation forecast information generated by the solar power generation management unit 228 and a reference forecast value indicating a standard solar power generation amount in the prediction target area. In the present disclosure, when there is no need to particularly distinguish between "solar radiation fluctuation information" and "solar power generation fluctuation information," they are collectively referred to as "fluctuation information." Details of the process of calculating the fluctuation information will be described later, and therefore will not be described here.

[0056] Next, in step S330, the fluctuation management unit 229 creates a GUI (Graphical User Interface) that displays the fluctuation amount information calculated in step S328. In one embodiment, the fluctuation management unit 229 may create a GUI that reflects the fluctuation amount information for each power plant calculated in step S328 and information on risk metrics that affect the amount of solar power generation on a map of the prediction target area specified in step S302. Note that an example of this GUI will be described with reference to FIG. 8, and therefore its description will be omitted here.

[0057] Next, in step S332, the fluctuation management unit 229 outputs the GUI generated in step S330. Here, the fluctuation management unit 229 may output the GUI via a display of the user terminal 260 or the like.

[0058] In the solar power generation management process 300 described above, by selecting an appropriate solar radiation prediction model and prediction information (past weather information, weather forecast information) according to the prediction time range desired by the user, it is possible to generate highly accurate solar radiation or solar power generation predictions for a wide area for any time range, whether short-term, medium-term, or long-term.

[0059] As described above, one aspect of an embodiment of the present disclosure relates to generating highly accurate solar radiation forecasts for a wide area for any time range, including short-term, medium-term, and long-term. To achieve this, the model training unit 222 according to an embodiment of the present disclosure preferably trains multiple solar radiation forecast models that generate solar radiation forecasts for different time ranges (short-term, medium-term, and long-term). Different learning information is used to train the solar radiation forecast models depending on the time range. Next, a process for training a solar radiation forecast model according to an embodiment of the present disclosure will be described with reference to FIGS. 4 and 5 .

[0060] 4 is a diagram illustrating an example of a process 400 for training a first solar radiation prediction model according to an embodiment of the present disclosure. The first solar radiation prediction model here is a machine learning model that predicts solar radiation for a first time range. Although the "first time range" here may be any time range, for convenience of explanation, a case where the "first time range" is a long period of time, such as three months or more, will be described as an example.

[0061] First, in step S402, the model training unit 222 acquires global solar radiation information 404 measured in a specific region (e.g., a first region) for a specific time range from the global solar radiation information DB 239. The "specific time range" here refers to a time range for which solar radiation prediction is desired, e.g., "four months" here. The model training unit 222 may perform a predetermined preprocessing on the acquired global solar radiation information 404 and convert the global solar radiation information 404 into a specific time interval (e.g., one day, one week, one month).

[0062] Also, in step S406, the model training unit 222 acquires from the past weather information DB 236 past weather information for learning 408 (first past weather information for learning), which is past weather information characterizing the climate of a specific region (e.g., a first region) for the same time range as in step S402. The model training unit 222 may perform predetermined preprocessing on the acquired past weather information for learning 408 to convert the past weather information for learning 408 into a specific time interval (e.g., one day, one week, one month, etc.). As described above, in principle, when predicting solar radiation for a longer time range (e.g., three months or more), it is desirable to use past weather information for a longer time range (in other words, past weather information that meets a predetermined length criterion). Using past weather information for a longer time range reduces errors and enables more accurate prediction results. Therefore, here, it is desirable for the model training unit 222 to acquire past weather information that meets a predetermined length criterion as the past weather information for learning 408. The long / short attribute may be defined for a time range for which a solar radiation prediction is desired. For example, if the time range for which a solar radiation prediction is desired is "four months," the long / short criterion may be "24 months or more," which is six times "four months."

[0063] Next, in step S410, the model training unit 222 uses the global solar radiation information 404 acquired in step S402 as ground truth 409 to train a machine learning model based on the learning past weather information 408. The machine learning model here may be, for example, a neural network, a long-short-term memory (LSTM) model, or Facebook Prophet (registered trademark). However, it is desirable that the machine learning model used here be a model that can capture periodic features that appear in long-term time-series information. This makes it possible to obtain a machine learning model that can generate predictions that take into account, for example, seasonal changes.

[0064] According to the prediction model training process 400 described above, a machine learning model that can capture periodic features that appear in long-term time series information is trained based on global solar radiation information and past weather information over a long time range that satisfies specified long-term and short-term attributes, thereby obtaining a solar radiation prediction model that can predict solar radiation with high accuracy over a long period of time.

[0065] 5 illustrates an example process 500 for training a second solar radiation prediction model according to an embodiment of the present disclosure. The second solar radiation prediction model here is a machine learning model that predicts solar radiation for a second time range. While the "second time range" here may be any time range, for ease of explanation, a case where the "second time range" is a short- to medium-term period of less than three months will be described as an example.

[0066] First, in step S502, the model training unit 222 acquires global solar radiation information 504 measured in a specific region (e.g., a second region) for a specific time range from the global solar radiation information DB 239. The "specific time range" here refers to a time range for which solar radiation prediction is desired, e.g., "two months" here. The model training unit 222 may perform a predetermined preprocessing on the acquired global solar radiation information 504 and convert the global solar radiation information 504 into a specific time interval (e.g., one day, one week, one month).

[0067] In step S506, the model training unit 222 acquires, from the past weather information DB 236, training past weather information 508 (second training past weather information), which is past weather information characterizing the climate of a specific region (e.g., a second region) for the same time range as in step S402. The model training unit 222 may perform predetermined preprocessing on the acquired training past weather information 508 to convert the training past weather information 508 into a specific time interval (one day, one week, one month, etc.). Note that when training a short- to medium-term solar radiation prediction model, it is not necessary to use past weather information for a longer time range than in a long-term solar radiation prediction model. Therefore, the acquired training past weather information 508 does not need to satisfy the long-short attribute described above. However, it goes without saying that it is desirable to use past weather information for a longer time range in order to reduce errors.

[0068] Also, in step S510, the model training unit 222 acquires, from the weather forecast information DB 238, training weather forecast information 512 indicating a weather forecast for a specific region (e.g., a first region) for the same time range as in step S402. The model training unit 222 may perform predetermined preprocessing on the acquired training weather forecast information 512 to convert the training weather forecast information 512 into a specific time interval (e.g., one day, one week, one month, etc.). As described above, the reason why the training weather forecast information 512 is used when training a short- to medium-term solar radiation prediction model is that, although the reliability of weather forecasts is limited for long-term periods, the reliability of weather forecasts is sufficiently high for short- to medium-term periods. Therefore, by predicting solar radiation based on weather forecast information in addition to the training historical weather information 508, more accurate prediction results can be obtained.

[0069] Next, in step S516, the model training unit 222 uses the global solar radiation information 504 acquired in step S502 as ground truth 514 to train a machine learning model based on the learning past weather information 508 and the learning weather forecast information 512. The machine learning model here may be, for example, a random forest regression model, but is not particularly limited to this.

[0070] According to the prediction model training process 500 described above, by training the machine learning model based on global solar radiation information, past weather information, and weather forecast information, it is possible to obtain a solar radiation prediction model that can predict solar radiation with high accuracy for short and medium terms.

[0071] Next, a solar power generation amount management unit according to an embodiment of the present disclosure will be described with reference to FIG. 6 .

[0072] 6 is a diagram illustrating an example of the operation of the solar power generation amount management unit 228 according to an embodiment of the present disclosure. As described above, the solar power generation amount management unit 228 according to an embodiment of the present disclosure can convert the solar radiation amount prediction information generated by the prediction unit 226 into solar power generation amount information indicating the predicted amount of solar power generation. The process of converting the solar radiation amount prediction information into solar power generation amount information is performed, for example, in step S322 of the solar power generation management process 300 shown in FIG. 3 described above.

[0073] More specifically, in an embodiment, the solar power generation amount management unit 228 can obtain a power generation conversion model 608 that converts solar radiation prediction information into solar power generation amount information by, for example, training a random forest regression model using global solar radiation information actually measured in a specific solar power plant, outputs 604 of multiple inverters in the solar power plant (i.e., solar power generation amount), and solar power generation facility information 606 indicating the relationship between the solar radiation amount and the inverter output (e.g., Supervisory Control and Data Acquisition data; SCADA).

[0074] Thereafter, the photovoltaic power generation amount management unit 228 can generate photovoltaic power generation amount forecast information 615 indicating the amount of photovoltaic power generation predicted at a specific power plant by analyzing the solar radiation forecast information 610 generated by the prediction unit 226 and the photovoltaic power generation facility information 606 indicating the relationship between the amount of solar radiation and the inverter output using the trained power generation conversion model 608. As described above, this photovoltaic power generation amount forecast information 615 may be output via a GUI indicating the status of photovoltaic power generation in a specified prediction target area. In addition, in some embodiments, the photovoltaic power generation amount management unit 228 may create information indicating the relationship between the solar radiation forecast information 610 and the inverter output in a graph format and output it via the GUI.

[0075] As described above, according to the photovoltaic power generation amount management unit 228 according to an embodiment of the present disclosure, by using global solar radiation information actually measured at a specific photovoltaic power plant, the output of multiple inverters at the photovoltaic power plant, and photovoltaic power generation equipment information indicating the relationship between the amount of solar radiation and the output of the inverters, for example, a random forest regression model can be trained to obtain a power generation amount conversion model that converts solar radiation prediction information into photovoltaic power generation amount information.

[0076] Next, a variation amount information generation process according to an embodiment of the present disclosure will be described with reference to FIG.

[0077] Fig. 7 is a diagram illustrating an example of the flow of a fluctuation amount information generation process 700 according to an embodiment of the present disclosure. The fluctuation amount information generation process 700 is a process for generating fluctuation amount information indicating a fluctuation amount that may occur in the predicted amount of solar radiation or the predicted amount of solar power generation, and is performed by the fluctuation management unit 229. Furthermore, the fluctuation amount information generation process 700 illustrated in Fig. 7 may be performed, for example, in step S328 of the solar power generation management process 300 illustrated in Fig. 3. It should be noted that the fluctuation amount information generation process 700 illustrated in Fig. 7 may be used to calculate the amount of fluctuation in the predicted amount of solar radiation or the amount of fluctuation in the amount of solar power generation.

[0078] First, in step S702, the fluctuation management unit 229 selects a power plant for which fluctuation information (solar radiation fluctuation information or solar power generation information) has not yet been generated from the prediction target area specified in step S302 of the solar power generation management process 300 shown in Figure 3.

[0079] Next, in step S704, the fluctuation management unit 229 acquires, for the power plant selected in step S702, the solar radiation forecast information or the solar power generation amount information generated in step S324 of the solar power generation management process 300 shown in Fig. 3 from the storage unit 230, and calculates a 95% confidence interval for the solar radiation forecast information or the solar power generation amount information. The 95% confidence interval is a statistical method that provides a range for estimating a parameter (e.g., an average or a percentage) of a population, and means that there is a 95% probability that the true parameter of the population lies within that interval.

[0080] In step S706, the fluctuation management unit 229 acquires the solar radiation forecast information or solar power generation amount information generated in step S324 of the solar power generation management process 300 shown in FIG. 3 for the power plant selected in step S702 from the storage unit 230 and calculates a reference forecast value for the solar radiation forecast information or solar power generation amount information. The reference forecast value here refers to information indicating a standard solar radiation or solar power generation amount calculated using an existing solar radiation or solar power generation amount calculation method, etc. In some embodiments, the fluctuation management unit 229 may calculate the reference forecast value by multiplying the maximum power generation amount or solar radiation of a specific power plant by a loss coefficient indicating past fluctuations or seasonal fluctuations. Note that a graph 725 shown in FIG. 7 graphically illustrates an example of the relationship between the solar radiation forecast information or solar power generation amount forecast information generated in step S324 of the solar power generation management process 300 shown in FIG. 3 and the reference forecast value.

[0081] Next, in step S708, the fluctuation management unit 229 calculates a fluctuation amount for the power plant selected in step S702 based on the solar radiation information or solar power generation amount information generated for the power plant and the reference forecast value calculated in step S706. More specifically, the fluctuation management unit 229 may calculate the fluctuation amount by subtracting the reference forecast value calculated in step S706 from the forecast solar radiation indicated in the solar radiation forecast information or the forecast solar power generation amount indicated in the solar power generation amount information. As an example, if the forecast value of solar power generation amount indicated in the solar power generation amount information is "100 MW" and the reference forecast value of solar power generation amount calculated in step S706 is "80 MW," the fluctuation amount may be calculated as "20 MW." Note that this fluctuation amount may be calculated for any period, such as daily, weekly, monthly, or annually.

[0082] Next, in step S710, the fluctuation management unit 229 calculates the standard deviation of the fluctuation amount calculated in step S708 for a short period of time, such as one day or one week.

[0083] Next, in step S712, the fluctuation management unit 229 calculates the standard deviation of the fluctuation amount calculated in step S708 for a medium-term or long-term period, such as one week, one month, or one year.

[0084] Next, in step S714, the fluctuation management unit 229 calculates the error in the amount of fluctuation based on the 95% confidence interval of the solar radiation amount forecast information or solar power generation amount information calculated in step S704 and the standard deviation of the amount of fluctuation calculated in steps S710 and S712. More specifically, the fluctuation management unit 229 may determine the error in the amount of fluctuation to be the sum of the 95% confidence interval of the solar radiation amount forecast information or solar power generation amount information calculated in step S704 and the standard deviation of the amount of fluctuation calculated in steps S710 and S712.

[0085] Next, in step S716, the fluctuation management unit 229 stores the fluctuation calculated in step S708 and the error calculated in step S714 for the power plant selected in step S702 in the storage unit 230. As described above, the fluctuation amount information generation process 700 may be used to calculate the fluctuation amount in the predicted solar radiation amount or the fluctuation amount in the amount of photovoltaic power generation. ３ (watts per cubic meter) or MW (megawatts), etc. Graph 750 shown in Fig. 7 graphically illustrates the fluctuation and error in the amount of solar power generation over a predetermined time range.

[0086] The fluctuation information generation process 700 described above can generate fluctuation information indicating the amount of fluctuation that may occur in the predicted amount of solar radiation or the predicted amount of solar power generation. By understanding the amount of fluctuation in solar radiation or the amount of solar power generation that may occur in the prediction target area, the amount of power supplied to the power grid can be predicted more accurately, making it possible to avoid power shortages and surplus power due to instability in solar power generation.

[0087] Next, a user interface screen according to an embodiment of the present disclosure will be described with reference to FIG.

[0088] Fig. 8 is a diagram illustrating an example of a user interface screen 800 according to an embodiment of the present disclosure. The user interface screen 800 illustrated in Fig. 8 is an example of a GUI for displaying information generated by the solar power generation management device 210 according to an embodiment of the present disclosure.

[0089] As shown in FIG. 8, a user interface screen 800 mainly includes a map display window 810 that displays a map of the prediction target area, and a photovoltaic power generation management window 820 that shows the status of photovoltaic power generation in the prediction target area.

[0090] The map display window 810 is an area that displays a map of the prediction target area. As shown in Fig. 8 , the map display window 810 displays each power plant (Site 1, Site 2, Site 3, Site 4) located in the prediction target area. By selecting a specific power plant in the map display window 810, the user can check detailed information about photovoltaic power generation at the selected power plant in the photovoltaic power generation management window 820.

[0091] Information showing the status of solar power generation in the prediction target area is displayed in the solar power generation management window 820. More specifically, the solar power generation management window 820 may include a region information display area 824 showing information about the entire prediction target area, and an individual information display area 828 showing information about individual power plants.

[0092] The regional information display area 824 may display, for example, the total power generation amount or the total fluctuation amount (fluctuation amount of solar radiation or fluctuation amount of photovoltaic power generation) for the entire region to be predicted. The individual information display area 828 may include, for each individual power plant, a risk metric indicating fluctuations in solar radiation or photovoltaic power generation amount, information on fluctuation amount of solar radiation, information on fluctuation amount of photovoltaic power generation amount, the reliability of the prediction, the contribution rate to the entire region, etc. The risk metric here may include, for example, the fluctuation amount from the previous year, the reliability of the prediction, the failure rate of power generation equipment, etc.

[0093] 8, the user can easily check the status of solar power generation in the target prediction area. By understanding the fluctuations in solar radiation and solar power generation that may occur in the target prediction area, the amount of power supplied to the power grid can be more accurately predicted, making it possible to avoid power shortages and surplus power due to instability in solar power generation.

[0094] As described above, one aspect of the solar power generation management means according to an embodiment of the present disclosure relates to selecting an appropriate solar radiation prediction model from among a plurality of pre-trained machine learning models according to a prediction time range specified by a user, and performing solar radiation prediction using the selected solar radiation prediction model. This allows for the generation of highly accurate solar radiation prediction results over a wide area for any short-term, medium-term, or long-term time range.

[0095] More specifically, for a long time range, for example, three months or more, a prediction is made using a first solar radiation prediction model trained on learning past weather information for a time range that satisfies a predetermined length criterion. By using the first solar radiation prediction model trained on learning past weather information for a time range that satisfies the predetermined length criterion, a prediction result can be obtained that takes into account the influence on solar radiation of periodic features that appear over a long period, such as seasonal changes.

[0096] Furthermore, for short-term and medium-term time ranges of, for example, less than three months, predictions are made using a second solar radiation prediction model trained on the learning weather forecast information in addition to the learning past weather information. Although the reliability of weather forecasts is limited for long-term periods, the reliability of weather forecasts is sufficiently high for short-term and medium periods, and by predicting solar radiation based on weather forecast information in addition to past weather information, more accurate prediction results can be obtained.

[0097] Furthermore, by using the power generation conversion model according to the embodiment of the present disclosure, it is possible to convert solar radiation prediction information generated by the solar radiation prediction model into solar power generation prediction information indicating the expected solar power generation amount, thereby making it possible to predict not only the solar radiation amount but also the expected solar power generation amount for each power plant.

[0098] Furthermore, according to the fluctuation information generation process of the embodiment of the present disclosure, it is possible to generate fluctuation information indicating the amount of fluctuation that may occur in the predicted amount of solar radiation or the predicted amount of solar power generation. By understanding the amount of fluctuation in solar radiation or the amount of solar power generation that may occur in the prediction target area, it is possible to more accurately predict the amount of power supplied to the power grid, thereby making it possible to avoid power shortages and surplus power due to instability in solar power generation.

[0099] In this way, according to the solar power generation management means of the embodiment of the present disclosure, it is possible to provide a solar power generation management means that is capable of generating highly accurate solar radiation or solar power generation forecasts for a wide area for any time range, whether short-term, medium-term, or long-term.

[0100] As described above, the solar power generation management means according to the embodiments of the present disclosure includes the following aspects.

[0101] (Aspect 1) A photovoltaic power generation management device comprising: a processor; a memory; and a storage unit for storing past weather information characterizing a climate of a predetermined region, wherein the memory includes: a model training unit for training a first solar radiation prediction model for predicting solar radiation for a first time range and a second solar radiation prediction model for predicting solar radiation for a second time range; a forecast information management unit for acquiring forecast request information including prediction target area information that defines a predetermined prediction target area and prediction time range information that defines a time range of prediction, and acquiring from the past weather information for forecasting past weather information that characterizes a climate of the prediction target area based on the prediction target area information and the prediction time range information included in the forecast request information; a prediction unit that generates solar radiation forecast information indicating the amount of solar radiation predicted in the prediction target area for the time range of the prediction by analyzing the past weather information for forecast using the first solar radiation forecast model if the prediction time range information included in the prediction request information satisfies a predetermined time range threshold, and that generates the solar radiation forecast information indicating the amount of solar radiation predicted in the prediction target area for the time range of the prediction by analyzing the past weather information for forecast using the second solar radiation forecast model if the prediction time range information included in the prediction request information does not satisfy the predetermined time range threshold; and a fluctuation management unit that generates and outputs solar radiation fluctuation information indicating a fluctuation in the amount of solar radiation predicted based on the solar radiation forecast information.

[0102] (Aspect 2) The photovoltaic power generation management device according to Aspect 1, wherein the model training unit trains the first solar radiation prediction model by learning a predetermined machine learning model using global solar radiation information measured in a first region for a first period and first learning historical weather information relating to a time range that characterizes the climate of the first region and satisfies a predetermined long / short criterion.

[0103] (Aspect 3) The photovoltaic power generation management device according to Aspect 1 or 2, characterized in that the model training unit trains the second solar radiation prediction model by learning a predetermined machine learning model using global solar radiation information measured in a second region for a second period, second learning past weather information characterizing the climate of the second region, and learning weather forecast information for the second region.

[0104] (Aspect 4) The solar power generation management device according to any one of Aspects 1 to 3, characterized in that the forecast information management unit determines a reference time range indicating a time range of the past weather information for keeping the error in the forecast accuracy of the amount of solar radiation within a predetermined tolerance threshold based on the forecast time range information, and acquires information corresponding to the reference time range from the past weather information as the past weather information for forecasting.

[0105] (Aspect 5) The photovoltaic power generation management device according to any one of Aspects 1 to 4, characterized in that the prediction request information further includes photovoltaic power generation facility information that characterizes photovoltaic power generation facilities to be installed in the prediction target area, and the photovoltaic power generation management device further includes a photovoltaic power generation amount management unit that generates photovoltaic power generation amount prediction information that indicates a photovoltaic power generation amount predicted in the prediction target area based on the solar radiation prediction information and the photovoltaic power generation facility information.

[0106] (Aspect 6) The photovoltaic power generation management device according to Aspect 5, wherein the fluctuation management unit calculates a reference forecast value indicating a standard photovoltaic power generation amount in the prediction target area based on a maximum photovoltaic power generation amount in the prediction target area and a loss coefficient indicating a fluctuation amount in the past photovoltaic power generation amount in the prediction target area, and generates photovoltaic power generation amount fluctuation information indicating a fluctuation amount in the photovoltaic power generation amount forecast information based on the photovoltaic power generation amount forecast information and the reference forecast value.

[0107] (Aspect 7) The photovoltaic power generation management device according to Aspect 6, wherein the fluctuation management unit provides a user interface that shows the solar radiation amount forecast information, the solar radiation amount fluctuation information, the photovoltaic power generation amount forecast information, and the photovoltaic power generation amount fluctuation information on a map showing the prediction target area.

[0108] (Aspect 8) The photovoltaic power generation management device according to any one of aspects 1 to 7, wherein the first time range is three months or more, and the second time range is less than three months.

[0109] Although the embodiments of the present invention have been described above, the present invention is not limited to the above-described embodiments, and various modifications are possible within the scope of the gist of the present invention.

[0110] 150: Photovoltaic power generation management application, 210: Photovoltaic power generation management device, 220: Memory, 222: Model training unit, 224: Forecast information management unit, 226: Forecast unit, 228: Photovoltaic power generation amount management unit, 229: Fluctuation management unit, 230: Storage unit, 236: Past weather information DB, 238: Weather forecast information DB, 239: Global solar radiation information DB 239, 244: Processor, 246: Input / output unit, 250: Communication network, 260: User terminal

Claims

1. A photovoltaic power generation management device comprising: a processor; a memory; and a storage unit that stores past weather information characterizing the climate of a specified region and weather forecast information for the specified region, wherein the memory includes: a model training unit that trains a first solar radiation prediction model that predicts solar radiation for a first time range and a second solar radiation prediction model that predicts solar radiation for a second time range; a prediction information management unit that acquires prediction request information including prediction target area information that defines a specified prediction target region and prediction time range information that defines the time range of prediction; a prediction unit that, if the prediction time range information included in the prediction request information satisfies a predetermined time range threshold, uses the first solar radiation prediction model to analyze past weather information for prediction that characterizes a climate of the prediction target area, obtained from the past weather information, to generate solar radiation prediction information indicating the amount of solar radiation predicted in the prediction target area for the time range of the prediction; and, if the prediction time range information included in the prediction request information does not satisfy the predetermined time range threshold, uses the second solar radiation prediction model to analyze the past weather information for prediction and weather forecast information for prediction for the prediction target area, obtained from the weather forecast information, to generate the solar radiation prediction information indicating the amount of solar radiation predicted in the prediction target area for the time range of the prediction; and a fluctuation management unit that generates and outputs solar radiation fluctuation information indicating a fluctuation in the amount of solar radiation predicted based on the solar radiation prediction information.

2. The solar power generation management device of claim 1, characterized in that the model training unit trains the first solar radiation prediction model by learning a predetermined machine learning model using global solar radiation information measured in a first region for a first period and first learning past weather information relating to a time range that characterizes the climate of the first region and meets predetermined long-short criteria.

3. The solar power generation management device of claim 1, characterized in that the model training unit trains the second solar radiation prediction model by learning a predetermined machine learning model using global solar radiation information measured in a second region for a second period, second learning past weather information characterizing the climate of the second region, and learning weather forecast information for the second region.

4. The solar power generation management device described in claim 1, characterized in that the forecast information management unit determines a reference time range indicating the time range of the past weather information for keeping the error in the forecast accuracy of solar radiation within a predetermined tolerance threshold based on the forecast time range information, and obtains information corresponding to the reference time range from the past weather information as the past weather information for forecasting.

5. The solar power generation management device according to claim 1, characterized in that the forecast request information further includes solar power generation facility information that characterizes solar power generation facilities to be installed in the forecast target area, and the solar power generation management device further includes a solar power generation amount management unit that generates solar power generation amount forecast information that indicates the amount of solar power generation predicted in the forecast target area based on the solar radiation forecast information and the solar power generation facility information.

6. The solar power generation management device according to claim 5, characterized in that the fluctuation management unit calculates a reference forecast value indicating a standard solar power generation amount in the prediction target area based on the maximum solar power generation amount in the prediction target area and a loss coefficient indicating the amount of fluctuation in the past solar power generation amount in the prediction target area, and generates solar power generation amount fluctuation information indicating the amount of fluctuation in the solar power generation amount forecast information based on the solar power generation amount forecast information and the reference forecast value.

7. The solar power generation management device described in claim 6, characterized in that the fluctuation management unit provides a user interface that shows the solar radiation forecast information, the solar radiation fluctuation information, the solar power generation forecast information, and the solar power generation fluctuation information on a map showing the prediction target area.

8. The photovoltaic power generation management device according to claim 1, characterized in that the first time range is three months or more, and the second time range is less than three months.

9. A photovoltaic power generation management system in which a photovoltaic power generation management device and a user terminal are connected via a communication network, the photovoltaic power generation management device comprising: a processor; a memory; and a storage unit for storing past weather information characterizing the climate of a specified region and weather forecast information for the specified region, the memory comprising: a model training unit for training a first solar radiation prediction model for predicting solar radiation for a first time range and a second solar radiation prediction model for predicting solar radiation for a second time range; a prediction information management unit for acquiring prediction request information including prediction target area information defining a specified prediction target region and prediction time range information defining the time range of the prediction; a prediction unit that, if the prediction time range information included in the prediction request information satisfies a predetermined time range threshold, uses the first solar radiation prediction model to analyze past weather information for prediction that characterizes a climate of the prediction target area, obtained from the past weather information, to generate solar radiation forecast information indicating the amount of solar radiation predicted in the prediction target area for the time range of the prediction; and, if the prediction time range information included in the prediction request information does not satisfy the predetermined time range threshold, uses the second solar radiation prediction model to analyze the past weather information for prediction and weather forecast information for prediction regarding the prediction target area, obtained from the weather forecast information, to generate the solar radiation forecast information indicating the amount of solar radiation predicted in the prediction target area for the time range of the prediction; and a fluctuation management unit that generates and outputs solar radiation fluctuation information indicating a fluctuation in the amount of solar radiation predicted based on the solar radiation forecast information.

10. A photovoltaic power generation management method executed by a photovoltaic power generation management device, the photovoltaic power generation management device comprising: a processor; a memory; and a storage unit that stores past weather information characterizing a climate of a predetermined region and weather forecast information for the predetermined region, the memory including: a step of training a first solar radiation prediction model that predicts solar radiation for a first time range by training a predetermined machine learning model using global solar radiation information measured in a first region for a first period and first learning past weather information that characterizes the climate of the first region and relates to a time range that satisfies a predetermined long / short criterion; and a step of training a second solar radiation prediction model that predicts solar radiation for a second time range by training a predetermined machine learning model using global solar radiation information measured in a second region for a second period, second learning past weather information that characterizes the climate of the second region, and learning weather forecast information for the second region. acquiring prediction request information including prediction target area information that defines a predetermined prediction target area and prediction time range information that defines a time range of prediction; if the prediction time range information included in the prediction request information satisfies a predetermined time range threshold, using the first solar radiation prediction model to analyze past weather information for prediction that characterizes the climate of the prediction target area, obtained from the past weather information, to generate solar radiation prediction information that indicates the amount of solar radiation predicted in the prediction target area for the time range of the prediction; if the prediction time range information included in the prediction request information does not satisfy the predetermined time range threshold, using the second solar radiation prediction model to analyze the past weather information for prediction and weather forecast information for prediction regarding the prediction target area, obtained from the weather forecast information, to generate the solar radiation prediction information that indicates the amount of solar radiation predicted in the prediction target area for the time range of the prediction; acquiring solar power generation facility information that characterizes solar power generation facilities to be installed in the prediction target area; generating solar power generation amount prediction information indicating a solar power generation amount predicted in the prediction target area based on the solar radiation amount prediction information and the solar power generation facility information;a step of calculating a reference forecast value indicating a standard amount of solar power generation in the prediction target area, based on a maximum amount of solar power generation in the prediction target area and a loss coefficient indicating a fluctuation amount of past solar power generation in the prediction target area; and a step of generating and outputting solar power generation fluctuation information indicating a fluctuation amount in the solar power generation forecast information, based on the solar power generation forecast information and the reference forecast value.

Images