Solar radiation amount data processing device and method

Abstract

A solar radiation amount data processing device estimates, on the basis of past power generation result data and past weather data from at least one first site where a photovoltaic power generation facility is established, a solar radiation amount estimation value which is the past solar radiation amount at the first site, and generates, on the basis of the past solar radiation amount estimation value and the past weather data at the first site, an interpolation model for outputting a solar radiation amount interpolation value obtained by interpolating the solar radiation amount from at least one second site where no photovoltaic power generation facility is established.

Description

Solar radiation data processing device and method

[0001] The disclosed technologies relate to a solar radiation data processing device and a solar radiation data processing method.

[0002] In the power system, a simultaneous supply and demand matching system is in place to stabilize the supply and demand of electricity, and power generators are required to minimize the difference between planned power generation and actual power generation. Photovoltaic (PV) power generation is easily affected by weather conditions, making it difficult to generate power according to the plan. Therefore, power generators are required to predict PV output in advance and formulate an accurate power generation plan.

[0003] Because PV output is strongly correlated with solar radiation, accurate solar radiation forecasts allow for accurate prediction of PV output. However, solar radiation forecasts using numerical weather prediction models, which are commonly based on weather satellite observation data, generally have large prediction errors. Even solar radiation forecasts made just before the start of actual demand do not have zero error, resulting in considerable errors in power generation plans.

[0004] One technique for reducing prediction errors in solar radiation is to combine physical methods with AI (Artificial Intelligence) to predict solar radiation while considering the behavior of the atmosphere and clouds (Non-Patent Literature 1).

[0005] Takeshi Utsunomiya, Jun Sasaki, Makoto Okada, Shigeyuki Yoshikawa, and Koji Yamaguchi, "Development of Short-Term Solar Radiation Prediction Technology Combining Physical Methods and AI," Proceedings of the 2023 Annual Convention of the Institute of Electrical Engineers of Japan, 6-242, pp. 420-421, 2023.

[0006] Conventional solar radiation forecasting techniques aim to improve the accuracy of solar radiation predictions by considering the movement, formation, and dissipation of clouds and other obstructions that affect solar radiation, as well as physical atmospheric processes, through observations from various sensors and image analysis of satellite images. However, a more fundamental cause of errors in solar radiation predictions is the accuracy of observations by weather satellites. Weather satellites, which are mainly used in conventional methods, have difficulty accurately observing the ground surface, and cannot accurately observe weather conditions near the ground, including solar radiation. This is a problem that causes errors in solar radiation predictions in various locations.

[0007] Furthermore, while there are technologies to estimate solar radiation equivalent to ground observations from power generation data, these technologies cannot estimate solar radiation at locations without PV facilities. Therefore, there is a problem in that it is not possible to estimate comprehensive, wide-area solar radiation based on the power generation data of PV facilities.

[0008] The disclosed technology was developed in light of the above points and aims to output comprehensive, wide-area solar radiation estimation data that takes into account meteorological conditions near the ground.

[0009] A first aspect of this disclosure is a solar radiation data processing device, comprising: a first estimation unit that estimates a solar radiation estimate, which is the past solar radiation at one or more first locations where solar power generation equipment is installed, based on past power generation performance data and past weather data at the first locations; and an interpolation learning unit that generates an interpolation model that outputs an interpolated solar radiation value obtained by interpolating the solar radiation at one or more second locations where solar power generation equipment is not installed, based on the past solar radiation estimate at the first locations and past weather data.

[0010] A second aspect of the present disclosure is a solar radiation data processing method performed by a solar radiation data processing device including a first estimation unit and an interpolation learning unit, wherein the first estimation unit estimates a solar radiation estimate, which is the past solar radiation at one or more first locations where solar power generation equipment is installed, based on past power generation performance data and past weather data at the first locations, and the interpolation learning unit generates an interpolation model that outputs an interpolated solar radiation value obtained by interpolating the solar radiation at one or more second locations where solar power generation equipment is not installed, based on the past solar radiation estimate at the first locations and past weather data.

[0011] According to the disclosed technology, it is possible to output comprehensive, wide-area solar radiation estimation data that takes into account weather conditions near the ground.

[0012] This is a block diagram showing the hardware configuration of the solar radiation data processing device. This is a functional block diagram of the solar radiation data processing device. This is a diagram explaining solar radiation estimates. This is a diagram explaining solar radiation interpolation values. This is a diagram explaining the data input range to the prediction model. This is a diagram explaining predictions made by the prediction model. This is a diagram explaining relative coordinates. This is a diagram showing an example of explanatory variables. This is a flowchart showing an example of machine learning processing. This is a flowchart showing an example of prediction processing.

[0013] An example of an embodiment of the disclosed technology will be described below with reference to the drawings. In each drawing, identical or equivalent components and parts are given the same reference numerals. Furthermore, the dimensional ratios in the drawings are exaggerated for illustrative purposes and may differ from actual ratios.

[0014] Figure 1 is a block diagram showing the hardware configuration of the solar radiation data processing device 10. As shown in Figure 1, the solar radiation data processing device 10 includes a CPU (Central Processing Unit) 11, a ROM (Read Only Memory) 12, a RAM (Random Access Memory) 13, storage 14, an input unit 15, a display unit 16, and a communication interface 17. Each component is connected to the others via a bus 19 so as to be able to communicate with each other.

[0015] The CPU 11 is a central processing unit that executes various programs and controls various parts. Specifically, the CPU 11 reads a program from the ROM 12 or storage 14 and executes the program using the RAM 13 as a working area. The CPU 11 controls each of the above components and performs various calculations according to the program stored in the ROM 12 or storage 14. In this embodiment, the ROM 12 or storage 14 stores a solar radiation data processing program, which will be described later.

[0016] ROM 12 stores various programs and data. RAM 13 temporarily stores programs or data as a working area. Storage 14 consists of storage devices such as HDD (Hard Disk Drive) and SSD (Solid State Drive) and stores various programs and data, including the operating system.

[0017] The input unit 15 includes a pointing device such as a mouse and a keyboard, and is used for various types of input. The display unit 16 is, for example, a liquid crystal display and displays various types of information. The display unit 16 may also function as the input unit 15 by employing a touch panel system.

[0018] Communication I / F17 is an interface for communicating with other devices. For this communication, wired communication standards such as Ethernet® and FDDI, or wireless communication standards such as 4G, 5G, and Wi-Fi® may be used.

[0019] Next, the functional configuration of the solar radiation data processing device 10 will be described. Figure 2 is a block diagram showing an example of the functional configuration of the solar radiation data processing device 10. As shown in Figure 2, the solar radiation data processing device 10 includes a machine learning unit 20 and a solar radiation prediction unit 40 as its functional configuration. The machine learning unit 20 further includes a first estimation unit 21, an interpolation learning unit 22, and a prediction learning unit 23. The solar radiation prediction unit 40 further includes a second estimation unit 41, an interpolation unit 42, and a prediction unit 43. Each functional configuration is realized by the CPU 11 reading a solar radiation data processing program stored in the ROM 12 or storage 14, expanding it into the RAM 13, and executing it.

[0020] Furthermore, a predetermined storage area of ​​the solar radiation data processing device 10 stores power generation data 1 DB (database) 24, weather data 1 DB 25, weather data 2 DB 26, and weather data 3 DB 27. Also, a predetermined storage area stores power generation data 2 DB 44, weather data 4 DB 45, weather data 5 DB 46, and weather data 6 DB 47. Note that each DB is not limited to being stored inside the solar radiation data processing device 10, but may also be stored in an external device accessible from the solar radiation data processing device 10.

[0021] The power generation performance database 1DB24 stores power generation performance data, which are the actual power generation amounts for each PV facility at each historical time point. Each time point may be, for example, at 10-minute intervals. The power generation performance database 1DB24 also stores the specifications of each PV facility necessary for estimating solar radiation data, such as the PCS (Power Conditioner) capacity, and the location information of the PV facility. The location information may be the latitude and longitude of the location where the PV facility is installed, or it may be information indicating which of multiple meshes, which divide the geographic area at predetermined distance intervals (for example, 2 km), it belongs to. For example, the power generation amount at time t for the PV facility with identification number i is stored in P ｉ，ｔ Therefore, the power generation record 1DB24 contains (i, t, P ｉ，ｔ Multiple power generation performance data, represented by ), are stored.

[0022] Weather data 1DB25, weather data 2DB26, and weather data 3DB27 each store weather data for each location at various points in the past. Each point in time may be, for example, at 10-minute intervals. Weather data 1DB25 stores data such as temperature, precipitation, and humidity. Weather data 2DB26 stores data such as cloud cover, which is the percentage of the sky covered by clouds. Weather data 3DB27 stores data such as wind direction and wind speed. For example, the location information of point j is stored in L ｊ W is the value of the weather at time t at point j. ｊ，ｔ Therefore, each of weather data 1DB25, weather data 2DB26, and weather data 3DB27 contains (j, t, L ｊ , W ｉ，ｔMultiple weather data points represented by ) are stored.

[0023] Power Generation Data Database 2DB44 stores power generation data for each PV facility at each time point within a predetermined period, including the current time. Power Generation Data Database 1DB24 and Power Generation Data Database 2DB44 differ only in the period over which power generation data is stored; the data structure of the stored power generation data is the same. Alternatively, Power Generation Data Database 1DB24 and Power Generation Data Database 2DB44 may be used interchangeably as databases that store power generation data for each time point in the past from the current time. The most recent power generation data in Power Generation Data Database 2DB44 is updated in real time.

[0024] Weather data 4DB45 and weather data 5DB46 each store weather data for each location at each time in the most recent predetermined period including the current time. Weather data 6DB47 stores weather data for each location at each time in the most recent predetermined period including the current time, and weather forecast data for each location at each future time. Each time may be, for example, at 30-minute intervals. Weather data 4DB45 stores data such as temperature, precipitation, and humidity. Weather data 5DB46 stores data such as cloud cover. Weather data 6DB47 stores data such as wind direction and wind speed. Weather data 1DB25, weather data 2DB26, and weather data 3DB27 differ from weather data 4DB45, weather data 5DB46, and weather data 6DB47 only in the period for which weather data is stored; the data structure of the stored weather data is the same. The most recent weather data in each database, as well as the latest weather forecast data for the forecast time in weather data 6DB47, are updated in real time.

[0025] Next, we will explain each functional unit in detail. First, we will explain the machine learning unit 20.

[0026] The first estimation unit 21 estimates solar radiation values, which are past solar radiation amounts at one or more locations where PV equipment is installed (hereinafter referred to as "estimation locations"), based on past power generation performance data and past weather data at the estimation locations. An estimation location is an example of a "first location" in this disclosure.

[0027] Specifically, the first estimation unit 21 acquires past power generation data and specifications and location information for each PV facility from the power generation data database 24. The first estimation unit 21 also acquires weather data, mainly temperature, precipitation, and humidity, as past weather data from the weather data database 25. Using the acquired data as features, the first estimation unit 21 estimates the solar radiation amount at the same location and time for each PV facility, for example, by multiple regression analysis. Note that the method for estimating the solar radiation amount is not limited to multiple regression analysis, and existing methods for estimating solar radiation from power generation (for example, see Reference 2) may be applied.

[0028] [Reference 2] Ryo Moriwaki, Shinji Tsuzuki, Wataru Miyao, Yuhei Sasagata, Kai Kajifusa, "Estimation of Total Solar Radiation Using Photovoltaic Power Generation," Transactions of the Japan Society of Civil Engineers, Series B1 (Hydraulic Engineering), Vol. 71, No. 4, pp. I_421-I_426, 2015.

[0029] Furthermore, as shown in the upper part of Figure 3, the first estimation unit 21 stores estimated solar radiation values ​​for each mesh, which is created by dividing the geographical area of ​​the target for which solar radiation is to be predicted into predetermined distance intervals (for example, 2 km intervals). The mesh in which PV equipment exists (white mesh in the lower part of Figure 3) corresponds to the estimation point. If multiple PV equipment exists in the same mesh, the first estimation unit 21 statistically processes the estimated solar radiation values ​​of each PV equipment at the same time, such as by averaging, and uses this as the estimated solar radiation value for the estimation point corresponding to that mesh. For example, the estimated solar radiation value at time t in the estimation point corresponding to the mesh with identification number m is I ｍ，ｔ Therefore, each mesh has (m, t, I ｍ，ｔ The data represented by ) is stored. The first estimation unit 21 passes the estimated past solar radiation values ​​of the estimated location to the interpolation learning unit 22 and the prediction learning unit 23.

[0030] The interpolation learning unit 22 generates an interpolation model 31 that interpolates and outputs an insolation amount interpolation value, which is the past insolation amount at one or more points where no PV equipment is installed (hereinafter referred to as "interpolation points"), based on the past estimated insolation amounts at the estimated points and the past meteorological data. The interpolation points are an example of the "second point" in the present disclosure. As shown in the upper diagram of FIG. 4, the white mesh is the estimated point, and the estimated insolation amount estimated value is stored therein. The other meshes are meshes where no PV equipment exists, that is, interpolation points, and the insolation amount estimated value estimated by the first estimation unit 21 is not stored therein.

[0031] Specifically, the interpolation learning unit 22 receives the past estimated insolation amounts at the estimated points from the first estimation unit 21 and acquires mainly cloud cover data as the past meteorological data from the meteorological data 2DB 26. The interpolation learning unit 22 executes machine learning of the interpolation model No. 3, which is a machine learning model, using the acquired data as feature amounts, thereby generating a trained interpolation model No. While more specifically, when a certain estimated point is set as the interpolation target point, the past estimated insolation amount estimated by the first estimation unit 21 for that estimated point is used as the correct data. Then, the interpolation learning unit 6 updates the parameters of the interpolation model 31 so that the insolation amount interpolation value of the interpolation target point interpolated by the interpolation model 31 approaches the correct data when the feature amounts are input, thereby executing the machine learning of the interpolation model 31.

[0032] As a result, for meshes where the insolation amount estimated value has not been estimated by the first estimation unit 21 (meshes other than the white meshes), as shown in the lower diagram of FIG. 4, the insolation amount interpolation value is interpolated as an interpolation point (meshes with dots), and a wide-area insolation amount without missing values at each point (each mesh) can be obtained. Hereinafter, the insolation amount obtained by combining the insolation amount estimated value at each estimated point and the insolation amount interpolation value at each interpolation point is referred to as "wide-area insolation amount estimation data". The data structure of the wide-area insolation amount estimation data is, for example, (m, t, I ｍ，ｔ ) as in the above-described insolation amount estimated value. In the case of the insolation amount estimated value, only the identification number of the mesh corresponding to the estimated point where the PV equipment exists corresponds to "m", but in the case of the wide-area insolation amount estimation data, the identification numbers of all meshes correspond.

[0033] The predictive learning unit 23 generates a predictive model 32 that predicts solar radiation values, which are solar radiation amounts at future times, based on wide-area solar radiation estimation data and past weather data. Specifically, the predictive learning unit 23 obtains wide-area solar radiation estimation data by receiving past solar radiation estimation values ​​for estimation points from the first estimation unit 21 and past solar radiation interpolation values ​​for interpolation points, which are the output of the interpolation model 31. The predictive learning unit 23 also obtains wind direction and wind speed data, mainly as past weather data, from the weather data 3DB 27.

[0034] The prediction learning unit 23 generates a prediction model 32 based on wide-area solar radiation estimation data and past wind direction and wind speed data, for example, by multiple regression analysis. The ground truth data is solar radiation data from acquired data at a time later than the time of prediction (for example, approximately 1 to 10 hours ahead). The time of the ground truth data is changed depending on the length of the prediction period, but a prediction model 32 that predicts different periods together may be generated by specifying the prediction period as an explanatory variable. The prediction learning unit 23 changes the range in which data is input for prediction of the target location according to the prediction period. For example, if the prediction period is from the current time to 3 hours ahead, the prediction learning unit 23 assumes an average cloud movement speed of 60 km / hour and inputs wide-area solar radiation estimation data and meteorological data for locations within a radius of 180 km from the target location, as shown in Figure 5.

[0035] Furthermore, as shown in Figure 6, the prediction model 32 tracks the changes in wide-area solar radiation from the past to the present in correspondence with meteorological data, thereby predicting the future solar radiation of the target location (for example, several hours ahead). In Figure 6, the mesh with thick lines represents the target location, and the shaded mesh represents locations where the solar radiation is lower than the surrounding area due to the influence of clouds, etc. In order to enable the prediction model 32 to make the above predictions, the prediction learning unit 23 performs machine learning on the prediction model 32 using wide-area solar radiation estimation data, which is time-series data of wide-area solar radiation, and wind direction and wind speed data, which represent changes in weather, as features. In addition, the prediction learning unit 23 may use wind direction and wind speed data at a time in the future beyond the start time of the prediction period as features in order to explain how the weather will change in the future.

[0036] More specifically, as shown in Figure 7, the predictive learning unit 23 represents the surrounding meshes using relative coordinates from the prediction target point (the mesh with a thick border in Figure 7), and uses solar radiation data and meteorological data for each time point corresponding to each mesh as features (explanatory variables). The example in Figure 7 shows an example in which data from points corresponding to an M x N mesh centered on the prediction target point is input to the predictive model 32 as features.

[0037] Figure 8 shows an example of explanatory variables for the prediction model 32. In the example in Figure 8, the current time is denoted as time t in each row, and the data structure stores solar radiation, wind direction, and wind speed data for each relative coordinate and time. In the example in Figure 8, wind vectors, which represent wind direction and wind speed as vectors, are used as the data for wind direction and wind speed. Solar radiation [a] at 12:30 on 7 / 7 / 2024 Ｍ－１，２ , t], solar radiation [a Ｍ，１ , t-30], and wind vector [a Ｍ，Ｎ From the value of t-30 (dashed line in Figure 8), the amount of solar radiation lower than the surrounding area is predicted for the target location [a Ｍ／２，Ｎ／２ It can be seen that this is propagating to [ ]. By training the prediction model 32 with this kind of relationship, a prediction model 32 capable of predicting solar radiation several hours in advance is generated.

[0038] Next, the solar radiation prediction unit 40 will be described.

[0039] The second estimation unit 41 estimates the estimated value of the solar radiation amount for the most recent predetermined period at the estimation point based on the power generation performance data for the most recent predetermined period and the meteorological data for the most recent predetermined period. Since the specific estimation method of the estimated value of the solar radiation amount is the same as that of the first estimation unit 21, a detailed description thereof is omitted.

[0040] The interpolation unit 42 inputs the estimated value of the solar radiation amount for the most recent predetermined period at the estimation point estimated by the second estimation unit 41 and the meteorological data for the most recent predetermined period into the interpolation model 31, and obtains the interpolated value of the solar radiation amount for the most recent predetermined period at the interpolation point output from the interpolation model 1.

[0041] The prediction unit 43 inputs the wide-area solar radiation amount estimation data for the most recent predetermined period and the meteorological data for the most recent predetermined period into the prediction model 32, and obtains the predicted value of the solar radiation amount at each point at a time later than the most recent predetermined period output from the prediction model 32.

[0042] Next, the operation of the solar radiation amount data processing device 10 will be described. FIG. 9 is a flowchart showing the flow of machine learning processing by the solar radiation amount data processing device 10. FIG. 10 is a flowchart showing the flow of prediction processing by the solar radiation amount data processing device 10. The CPU 11 reads out the solar radiation amount data processing program from the ROM 12 or the storage 14, expands it in the RAM 13, and executes it, thereby performing machine learning processing and prediction processing.

[0043] First, the machine learning processing shown in FIG. 9 will be described in detail.

[0044] In step S11, the CPU 11 estimates the past estimated value of the solar radiation amount at the estimation point as the first estimation unit 21 based on the past power generation performance data and the past meteorological data at the estimation point.

[0045] Next, in step S12, the CPU 11 generates, as the interpolation learning unit 22, an interpolation model 31 that interpolates and outputs the past interpolated value of the solar radiation amount at the interpolation point based on the past estimated value of the solar radiation amount at the estimation point and the past meteorological data, and stores it in a predetermined storage area.

[0046] Next, in step S13, the CPU 11, acting as a predictive learning unit 23, acquires the past solar radiation prediction values ​​for the estimated location and the past solar radiation interpolation values ​​for the interpolated location as past wide-area solar radiation estimation data. Then, the CPU 11, acting as a predictive learning unit 23, generates a predictive model 32 that predicts the solar radiation prediction value for future times based on the past wide-area solar radiation estimation data and past weather data, and stores it in a predetermined memory area. The machine learning process then ends.

[0047] Next, we will describe in detail the prediction process shown in Figure 10.

[0048] In step S21, the CPU 11, as a second estimation unit 41, estimates the solar radiation estimate for the most recent predetermined period at the estimation point based on the power generation performance data and weather data for the most recent predetermined period.

[0049] Next, in step S22, the CPU 11, as the interpolation unit 42, inputs the estimated solar radiation values ​​for the most recent predetermined period at the estimation point, estimated by the second estimation unit 41, and the meteorological data for the most recent predetermined period to the interpolation model 31. Then, the CPU 11, as the interpolation unit 42, acquires the interpolated solar radiation values ​​for the most recent predetermined period at the interpolation point, which are output from the interpolation model 31.

[0050] Next, in step S23, the CPU 11, as the prediction unit 43, acquires the estimated solar radiation values ​​for the most recent predetermined period at the estimated location and the interpolated solar radiation values ​​for the most recent predetermined period at the interpolated location, as the wide-area solar radiation estimation data for the most recent predetermined period. Then, the CPU 11, as the prediction unit 43, inputs the wide-area solar radiation estimation data for the most recent predetermined period and the meteorological data for the most recent predetermined period into the prediction model 32, predicts the solar radiation values ​​for a time later than the predetermined period (future), and outputs the prediction result. The prediction process then ends.

[0051] As described above, the solar radiation data processing device according to this embodiment estimates past solar radiation values ​​at one or more estimated locations where PV equipment is installed, based on past power generation performance data and past weather data. The solar radiation data processing device also generates an interpolation model that outputs interpolated solar radiation values ​​at one or more interpolation locations where PV equipment is not installed, based on past solar radiation estimates at the estimated locations and past weather data. In this way, this embodiment utilizes the power generation amount of PV equipment, which has a strong correlation with ground-level solar radiation, and also interpolates locally changing solar radiation amounts from surrounding solar radiation estimates and weather data such as clouds in the upper atmosphere for locations where PV equipment is not installed, thereby enabling the output of wide-area solar radiation estimation data without gaps.

[0052] Furthermore, the solar radiation data processing device according to this embodiment generates a prediction model that predicts solar radiation values ​​for future times based on past solar radiation estimates at the estimation point, past solar radiation interpolated values ​​at the interpolation point interpolated by the interpolation model, and past meteorological data. By using this prediction model to predict solar radiation values ​​for future times, it is possible to predict the solar radiation at a target point from the time-series changes of wide-area solar radiation data equivalent to ground observations, which are difficult to observe from weather satellites, thereby enabling highly accurate solar radiation prediction.

[0053] Furthermore, when using the solar radiation interpolation values ​​obtained from the interpolation model as input data, the prediction model may be a statistical prediction model such as a machine learning model or a numerical weather prediction model using physical methods.

[0054] Furthermore, the machine learning and prediction processes that the CPU reads and executes in the above embodiment may be executed by various processors other than the CPU. Examples of such processors include PLDs (Programmable Logic Devices) whose circuit configuration can be changed after manufacturing, such as FPGAs (Field-Programmable Gate Arrays), and dedicated electrical circuits that have a circuit configuration specifically designed to execute a particular process, such as ASICs (Application Specific Integrated Circuits). The machine learning and prediction processes may be executed by one of these various processors, or by a combination of two or more processors of the same or different types (for example, multiple FPGAs, and a combination of a CPU and an FPGA). More specifically, the hardware structure of these various processors is an electrical circuit that combines circuit elements such as semiconductor elements.

[0055] Furthermore, although the above embodiment describes a configuration in which the solar radiation data processing program is pre-stored (installed) on storage, the program is not limited to this. The program may be provided in a form stored on a non-transitor storage medium such as a CD-ROM (Compact Disk Read Only Memory), DVD-ROM (Digital Versatile Disk Read Only Memory), or USB (Universal Serial Bus) memory. Alternatively, the program may be provided in a form that is downloaded from an external device via a network.

[0056] The following additional information is disclosed regarding the embodiments described above.

[0057] (Note 1) A solar radiation data processing device comprising: a first estimation unit that estimates a solar radiation estimate, which is the past solar radiation at one or more first locations where solar power generation equipment is installed, based on past power generation performance data and past weather data at the first location; and an interpolation learning unit that generates an interpolation model that outputs an interpolated solar radiation value obtained by interpolating the solar radiation at one or more second locations where solar power generation equipment is not installed, based on the past solar radiation estimate at the first location and past weather data.

[0058] (Appendix 2) The solar radiation data processing device according to Appendix 1, wherein the meteorological data used in the first estimation unit includes at least temperature, precipitation, and humidity, and the meteorological data used in the interpolation learning unit includes at least cloud cover.

[0059] (Appendix 3) A solar radiation data processing device according to Appendix 1 or Appendix 2, which includes a predictive learning unit that generates a predictive model for predicting solar radiation, which is the amount of solar radiation at a future time, based on past solar radiation estimates at the first location, past solar radiation interpolated values ​​at the second location interpolated by the interpolation model, and past weather data.

[0060] (Appendix 4) The weather data used in the predictive learning unit is the solar radiation data processing device described in Appendix 3, which includes at least wind direction and wind speed.

[0061] (Appendix 5) A solar radiation data processing device according to Appendix 3 or Appendix 4, comprising: a second estimation unit that estimates the solar radiation estimate for the first location for the predetermined period based on power generation performance data for the most recent predetermined period and weather data for the predetermined period; an interpolation unit that inputs the solar radiation estimate for the first location for the predetermined period and the weather data for the predetermined period into an interpolation model to interpolate the solar radiation interpolation value for the second location for the predetermined period; and a prediction unit that inputs the solar radiation estimate for the first location for the predetermined period, the solar radiation interpolation value for the second location for the predetermined period interpolated by the interpolation model, and the weather data for the predetermined period into a prediction model to predict the solar radiation prediction value for a time later than the predetermined period.

[0062] (Appendix 6) A solar radiation data processing program for causing a computer to function as one of the components of the solar radiation data processing device described in any one of Appendix 1 to Appendix 5.

[0063] (Appendix 7) A solar radiation data processing device comprising: a memory; and at least one processor connected to the memory, wherein the processor is configured to estimate solar radiation values, which are past solar radiation amounts at one or more first locations where solar power generation equipment is installed, based on past power generation performance data and past weather data at the first locations, and to generate an interpolation model that outputs an interpolated solar radiation value obtained by interpolating the solar radiation amounts at one or more second locations where solar power generation equipment is not installed, based on the past solar radiation estimates at the first locations and past weather data.

[0064] (Appendix 8) A non-temporary storage medium storing a program executable by a computer to perform solar radiation data processing, wherein the solar radiation data processing estimates solar radiation values, which are past solar radiation at one or more first locations where solar power generation equipment is installed, based on past power generation performance data and past weather data at the first locations, and generates an interpolation model that outputs an interpolated solar radiation value obtained by interpolating the solar radiation at one or more second locations where solar power generation equipment is not installed, based on the past solar radiation estimate at the first locations and past weather data.

[0065] 10 Solar radiation data processing unit 11 CPU 12 ROM 13 RAM 14 Storage 15 Input unit 16 Display unit 17 Communication interface 19 Bus 20 Machine learning unit 21 First estimation unit 22 Interpolation learning unit 23 Predictive learning unit 24 Power generation performance 1 DB 25 Weather data 1 DB 26 Weather data 2 DB 27 Weather data 3 DB 31 Interpolation model 32 Predictive model 40 Prediction unit 41 Second estimation unit 42 Interpolation unit 43 Prediction unit 44 Power generation performance 2 DB 45 Weather data 4 DB 46 Weather data 5 DB 47 Weather data 6 DB

Claims

1. A solar radiation data processing device comprising: a first estimation unit that estimates a solar radiation estimate, which is the past solar radiation at one or more first locations where solar power generation equipment is installed, based on past power generation performance data and past weather data at the said first location; and an interpolation learning unit that generates an interpolation model that outputs an interpolated solar radiation value obtained by interpolating the solar radiation at one or more second locations where solar power generation equipment is not installed, based on the past solar radiation estimate at the said first location and past weather data.

2. The solar radiation data processing device according to claim 1, comprising a predictive learning unit that generates a predictive model for predicting solar radiation, which is the amount of solar radiation at a future time, based on past solar radiation estimates at the first location, past solar radiation interpolated values ​​at the second location interpolated by the interpolation model, and past weather data.

3. A solar radiation data processing device according to claim 2, comprising: a second estimation unit that estimates the solar radiation estimate for the first location for the predetermined period based on power generation performance data for the most recent predetermined period and weather data for the predetermined period; an interpolation unit that inputs the solar radiation estimate for the first location for the predetermined period and the weather data for the predetermined period into an interpolation model to interpolate the solar radiation interpolation value for the second location for the predetermined period; and a prediction unit that inputs the solar radiation estimate for the first location for the predetermined period, the solar radiation interpolation value for the second location for the predetermined period interpolated by the interpolation model, and the weather data for the predetermined period into a prediction model to predict the solar radiation prediction value for a time later than the predetermined period.

4. A solar radiation data processing method performed by a solar radiation data processing device including a first estimation unit and an interpolation learning unit, wherein the first estimation unit estimates a solar radiation estimate, which is the past solar radiation at one or more first locations where solar power generation equipment is installed, based on past power generation performance data and past weather data at those locations, and the interpolation learning unit generates an interpolation model that outputs an interpolated solar radiation value obtained by interpolating the solar radiation at one or more second locations where solar power generation equipment is not installed, based on the past solar radiation estimate at the first locations and past weather data.

Images