Photovoltaic generating capacity prediction method and device based on multi-source data fusion

Abstract

The invention discloses a photovoltaic generating capacity prediction method and device based on multi-source data fusion, and relates to the technical field of photovoltaic power generation, and the method comprises the steps: obtaining the meteorological prediction data of a plurality of meteorological data sources at each time point in a to-be-predicted time range, and determining the target meteorological type of each time point; determining a target weight of each meteorological data source according to a first weight of each meteorological data source corresponding to the target meteorological type and a second weight of each meteorological data source corresponding to the predicted duration type of the to-be-predicted time range; according to the target weight of each meteorological data source, fusing the meteorological prediction data of the plurality of meteorological data sources at each time point in the to-be-predicted time range to obtain meteorological fusion data of each time point; and inputting the meteorological fusion data of each time point into a pre-trained photovoltaic power generation prediction model to obtain predicted generating capacity information of each time point in the to-be-predicted time range. According to the invention, the data accuracy of the data source can be improved, and the prediction precision of the photovoltaic generating capacity is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of photovoltaic power generation, and particularly relates to a multi-source data fusion photovoltaic power generation prediction method and device. BACKGROUND

[0002] With the global energy transformation and rapid development of renewable energy, photovoltaic power generation systems are widely used in residential homes, commercial buildings and industrial parks. However, photovoltaic power generation has obvious intermittency and volatility characteristics, and its power generation is affected by natural environment and other factors. This uncertainty brings challenges to the user's power planning and the power grid's dispatching management. Therefore, accurate prediction of photovoltaic power generation is particularly important.

[0003] The prior art mainly relies on meteorological data provided by a single meteorological data source for photovoltaic power generation prediction. However, due to the significant differences in data accuracy of a single meteorological data source in different regions and weather conditions, the prediction accuracy of photovoltaic power generation is low. SUMMARY

[0004] The embodiment of the present application provides a multi-source data fusion photovoltaic power generation prediction method to improve the data accuracy of the data source and further improve the prediction accuracy of the photovoltaic power generation. The method comprises: obtaining meteorological prediction data of a plurality of meteorological data sources at each time point in a to-be-predicted time range, and determining a target meteorological type of each time point; determining a target weight of each meteorological data source according to a first weight of each meteorological data source corresponding to the target meteorological type and a second weight of each meteorological data source corresponding to a prediction length type of the to-be-predicted time range; fusing the meteorological prediction data of the plurality of meteorological data sources at each time point in the to-be-predicted time range according to the target weight of each meteorological data source to obtain meteorological fusion data of each time point; inputting the meteorological fusion data of each time point into a pre-trained photovoltaic power generation prediction model to obtain prediction power generation information of each time point in the to-be-predicted time range, wherein the photovoltaic power generation prediction model is obtained by training a machine learning model using historical meteorological data and corresponding historical photovoltaic power generation information.

[0005] Preferably, before the target weight of each meteorological data source is determined according to the first weight of each meteorological data source corresponding to the target meteorological type and the second weight of each meteorological data source corresponding to the prediction length type of the to-be-predicted time range, the method further comprises: determining the first weight of each meteorological data source corresponding to the target meteorological type according to pre-set meteorological data source weight information corresponding to different meteorological types.

[0006] Preferably, before the target weight of each meteorological data source is determined according to the first weight of each meteorological data source corresponding to the target meteorological type and the second weight of each meteorological data source corresponding to the prediction length type of the to-be-predicted time range, the method further comprises the following steps of: determining the prediction length type according to the length information of the to-be-predicted time range; acquiring historical meteorological prediction data of each meteorological data source for the same prediction length type and corresponding historical real meteorological data in a preset historical time period; determining the second weight of each meteorological data source according to the deviation of the historical meteorological prediction data of each meteorological data source and the corresponding historical real meteorological data.

[0007] Preferably, the step of determining the second weight of each meteorological data source according to the deviation of the historical meteorological prediction data of each meteorological data source and the corresponding historical real meteorological data comprises the following steps of: for each meteorological data source: calculating error data of the historical meteorological prediction data of different data types and the corresponding historical real meteorological data at the same time point; and fusing the error data of different data types at the same time point according to a preset third weight corresponding to different data types to obtain comprehensive error of each time point; determining the second weight of each meteorological data source according to the comprehensive error of each time point of each meteorological data source.

[0008] Preferably, before the predicted power generation information at each time point in the to-be-predicted time range is obtained by inputting the meteorological fusion data of each time point into the pre-trained photovoltaic power generation prediction model, the method further comprises the following steps of: extracting features from the meteorological fusion data of each time point to obtain meteorological feature data of each time point, wherein the meteorological feature data comprises temperature features, cloud coverage features, solar radiation intensity features, meteorological type features, hour-periodic features, date-periodic features, and / or seasonal periodic features; the step of obtaining the predicted power generation information at each time point in the to-be-predicted time range by inputting the meteorological fusion data of each time point into the pre-trained photovoltaic power generation prediction model comprises the following step of: inputting the meteorological feature data of each time point into the pre-trained photovoltaic power generation prediction model to obtain the predicted power generation information at each time point in the to-be-predicted time range.

[0009] Preferably, the step of acquiring the meteorological prediction data of each time point in the to-be-predicted time range from the plurality of meteorological data sources comprises the following steps of: determining a data acquisition frequency according to the prediction length type of the to-be-predicted time range; According to the data acquisition frequency, meteorological prediction data in a to-be-predicted time range is acquired from a plurality of meteorological data sources, to obtain meteorological prediction data of each time point of each meteorological data source.

[0010] Preferably, the method further comprises: Acquiring solar position information and photovoltaic panel temperature information in the to-be-predicted time range. The method further comprises: The method further comprises:

[0011] The application further provides a multi-source data fusion photovoltaic power generation amount prediction device to improve the data accuracy of data sources and the prediction accuracy of photovoltaic power generation amount. The first data acquisition module is configured to acquire meteorological prediction data of each time point in a to-be-predicted time range of a plurality of meteorological data sources, and determine a target meteorological type of each time point. The weight determination module is configured to determine a target weight of each meteorological data source according to a first weight of each meteorological data source corresponding to the target meteorological type and a second weight of each meteorological data source corresponding to a prediction time length type of the to-be-predicted time range. The data fusion module is configured to fuse the meteorological prediction data of each time point in the to-be-predicted time range of the plurality of meteorological data sources according to the target weight of each meteorological data source, to obtain meteorological fusion data of each time point. The model prediction module is configured to input the meteorological fusion data of each time point into a pre-trained photovoltaic power generation prediction model, to obtain prediction power generation amount information of each time point in the to-be-predicted time range.

[0012] The application further provides a computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the multi-source data fusion photovoltaic power generation amount prediction method when executing the computer program.

[0013] The embodiment of the present application also provides a computer readable storage medium, which stores a computer program, and the computer program is executed by a processor to realize the photovoltaic power generation amount prediction method of multi-source data fusion.

[0014] The embodiment of the present application also provides a computer program product, which comprises a computer program, and the computer program is executed by a processor to realize the photovoltaic power generation amount prediction method of multi-source data fusion.

[0015] In the embodiment of the present application, the meteorological prediction data of each time point in the to-be-predicted time range of a plurality of meteorological data sources is acquired, and the target meteorological type of each time point is determined; the target weight of each meteorological data source is determined according to the first weight of each meteorological data source corresponding to the target meteorological type and the second weight of each meteorological data source corresponding to the prediction length type of the to-be-predicted time range; the meteorological prediction data of each time point in the to-be-predicted time range of the plurality of meteorological data sources is fused according to the target weight of each meteorological data source, to obtain the meteorological fusion data of each time point; and the meteorological fusion data of each time point is input into a pre-trained photovoltaic power generation prediction model to obtain the predicted power generation amount information of each time point in the to-be-predicted time range, wherein the photovoltaic power generation prediction model is obtained by training a machine learning model by using historical meteorological data and corresponding historical photovoltaic power generation amount information. Compared with the existing method of predicting photovoltaic power generation by relying on a single meteorological data source, the embodiment of the present application predicts the photovoltaic power generation amount by fusing the meteorological prediction data of a plurality of meteorological data sources, and dynamically sets the target weight of each meteorological data source in combination with the meteorological type and the prediction length type of the to-be-predicted time range, thereby ensuring the scientificity and accuracy of the weight distribution, truly reflecting the actual capability of each meteorological data source in different prediction scenarios, and further improving the prediction accuracy of the photovoltaic power generation amount. BRIEF DESCRIPTION OF DRAWINGS

[0016] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or the prior art description. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can also be obtained by those skilled in the art without any creative effort.

[0017] In the drawings: Figure 1 A flowchart of a photovoltaic power generation amount prediction method of multi-source data fusion provided by the embodiment of the present application is shown in the drawings. Figure 2 A flowchart of another photovoltaic power generation amount prediction method of multi-source data fusion provided by the embodiment of the present application is shown in the drawings. Figure 3A schematic diagram of a multi-source data fusion photovoltaic power generation amount prediction device provided for an embodiment of the present application is shown in the figure. Figure 4 A schematic diagram of another multi-source data fusion photovoltaic power generation amount prediction device provided for an embodiment of the present application is shown in the figure. Figure 5 A schematic diagram of a computer device in an embodiment of the present application is shown in the figure. DETAILED DESCRIPTION

[0018] To make the purpose, technical solutions and advantages of the embodiments of the present application clearer and more apparent, the embodiments of the present application are further described in detail below with reference to the drawings. Herein, the schematic embodiments of the present application and the descriptions thereof are used to explain the present application, but not as a limitation on the present application.

[0019] In the description of the present specification, “comprise”, “include”, “have”, “contain” and the like are all open terms, that is, they mean containing but not limited to. The description of the terms “one embodiment”, “one specific embodiment”, “some embodiments”, “for example” and the like means that the specific features, structures or characteristics described in connection with the embodiment or example are contained in at least one embodiment or example of the present application. In the present specification, the schematic description of the above terms does not necessarily mean the same embodiment or example. Moreover, the specific features, structures or characteristics described can be combined in any one or more embodiments or examples in a suitable manner. The order of the steps involved in the embodiments is used to illustrate the embodiments of the present application, and the order of the steps is not limited, and can be appropriately adjusted as needed.

[0020] It is found through research that the prior art mainly relies on a single meteorological data source for photovoltaic power generation prediction, but the accuracy of a single meteorological data source under different regions and weather conditions varies significantly. Specifically, it is shown that: 1. Regional difference problem: the coverage quality and accuracy of different meteorological data sources are inconsistent in different regions, and some meteorological data sources may have obvious deviations in specific regions; 2. Meteorological type adaptability problem: the prediction accuracy of a single meteorological data source significantly decreases under extreme or rapidly changing weather conditions (such as sudden changes from sunny to cloudy, local heavy rainfall, etc.); 3. Time scale (prediction length type) problem: the accuracy of a single meteorological data source varies in different prediction time ranges (short-term, medium-term, long-term), and it is difficult to simultaneously meet the prediction needs of different time scales; 4. Data update frequency limitation: the update frequency of a single meteorological data source is fixed, and it is difficult to meet the differentiated update needs of different time scale predictions.

[0021] Therefore, due to the significant difference in data accuracy of the meteorological prediction data under different prediction scenarios, the use of the meteorological prediction data to predict the photovoltaic power generation capacity results in low prediction accuracy of the photovoltaic power generation capacity.

[0022] Based on this, the embodiment of the present application provides a photovoltaic power generation capacity prediction scheme based on multi-source data fusion, which predicts the photovoltaic power generation capacity based on multi-source data fusion under different prediction scenarios, improves the data accuracy of the data source, and further improves the prediction accuracy of the photovoltaic power generation capacity.

[0023] Figure 1 A flowchart of a photovoltaic power generation capacity prediction method based on multi-source data fusion is provided for the embodiment of the present application. The execution subject of the method can be an electronic device, or a functional module or functional entity capable of realizing the photovoltaic power generation capacity prediction method based on multi-source data fusion in the electronic device, and the above-mentioned electronic device includes but is not limited to a mobile terminal, a tablet computer, a computer, a camera, a wearable device, etc.

[0024] As shown in Figure 1 The photovoltaic power generation capacity prediction method based on multi-source data fusion can include: Step 101, obtaining meteorological prediction data of a plurality of meteorological data sources at each time point in a to-be-predicted time range, and determining a target meteorological type of each time point; Step 102, determining a target weight of each meteorological data source according to a first weight of each meteorological data source corresponding to the target meteorological type and a second weight of each meteorological data source corresponding to a prediction time length type of the to-be-predicted time range; Step 103, fusing the meteorological prediction data of the plurality of meteorological data sources at each time point in the to-be-predicted time range according to the target weight of each meteorological data source, to obtain meteorological fusion data at each time point; Step 104, inputting the meteorological fusion data at each time point into a pre-trained photovoltaic power generation prediction model to obtain prediction power generation capacity information at each time point in the to-be-predicted time range, and the photovoltaic power generation prediction model is obtained by training a machine learning model using historical meteorological prediction data of a plurality of meteorological data sources and corresponding historical photovoltaic power generation capacity information.

[0025] In the embodiment of the present application, the meteorological prediction data of each time point in the to-be-predicted time range is obtained from a plurality of meteorological data sources, and the target meteorological type of each time point is determined; the target weight of each meteorological data source is determined according to the first weight of each meteorological data source corresponding to the target meteorological type and the second weight of each meteorological data source corresponding to the prediction length type of the to-be-predicted time range; the meteorological fusion data of each time point is obtained by fusing the meteorological prediction data of each time point in the to-be-predicted time range of the plurality of meteorological data sources according to the target weight of each meteorological data source; and the predicted power generation information of each time point in the to-be-predicted time range is obtained by inputting the meteorological fusion data of each time point into a pre-trained photovoltaic power generation prediction model, which is obtained by training a machine learning model using historical meteorological data and corresponding historical photovoltaic power generation information. Compared with the existing method of predicting photovoltaic power generation by relying on a single meteorological data source, the embodiment of the present application predicts the photovoltaic power generation by fusing the meteorological prediction data of a plurality of meteorological data sources, and dynamically sets the target weight of each meteorological data source in combination with the meteorological type and the prediction length type of the to-be-predicted time range, thereby ensuring the scientificity and accuracy of the weight distribution, truly reflecting the actual ability of each meteorological data source in different prediction scenarios, and further improving the prediction accuracy of photovoltaic power generation.

[0026] The above-mentioned photovoltaic power generation prediction method based on multi-source data fusion will be described below.

[0027] In step 101, the meteorological prediction data of each time point in the to-be-predicted time range can be obtained from a plurality of meteorological data sources, and the target meteorological type of each time point is determined.

[0028] In specific implementation, different meteorological data sources have different characteristics. For example, the first meteorological data source has the longest prediction range, can predict meteorological data for 16 days in the future, updates data every 3 hours, and has a wide coverage of prediction location, which is suitable for obtaining meteorological prediction data in a global range; the second meteorological data source has the longest prediction range of 15 days, has a prediction range of 5 days at the hour level (i.e., predicting meteorological data every hour in the next 5 days), updates data every 1 hour, has high local prediction accuracy, and is suitable for obtaining meteorological prediction data in a specific range; and the third meteorological data source has the longest prediction range of 10 days, has a prediction range of 7 days at the hour level (i.e., predicting meteorological data every hour in the next 7 days), updates data in real time, has strong real-time performance based on a ground meteorological station network, has accurate local data, and is suitable for obtaining real-time meteorological data and short-term forecast meteorological prediction data.

[0029] The meteorological prediction data at each time point can include weather data such as temperature, cloud cover, precipitation probability, wind speed, humidity, and solar radiation intensity at each time point. The weather type can be sunny, rainy, cloudy, snowy, extreme weather, and the like; and the target weather type at each time point can be determined by each meteorological data source according to the weather type of the meteorological prediction data at each time point, and then the target weather type at each time point can be obtained through a voting mechanism.

[0030] In one embodiment, the step 101 can specifically include: According to the prediction length type of the to-be-predicted time range, determining a data collection frequency; According to the data collection frequency, collecting meteorological prediction data in the to-be-predicted time range from the plurality of meteorological data sources to obtain meteorological prediction data at each time point of the plurality of meteorological data sources.

[0031] In specific implementation, the to-be-predicted time range can be a pre-set future time range, for example, 24 hours in the future, 3 days in the future, 14 days in the future, and the like. According to the prediction length type (or time scale type) of the to-be-predicted time range, different data collection frequencies (data collection time intervals) can be set, and the meteorological prediction data in the to-be-predicted time range can be collected from the plurality of meteorological data sources at the data collection time intervals, for example, if the to-be-predicted time range is 24 hours in the future, the prediction length type is considered to be short-term prediction, and the meteorological prediction data in the future 24 hours can be collected at an interval of one hour, and each time point is every one hour; if the to-be-predicted time range is 1-6 days in the future, the prediction length type is considered to be medium-term prediction, and the meteorological prediction data in the future 1-6 days can be collected at an interval of 6 hours, and each time point is every 6 hours; if the to-be-predicted time range is 7-14 days in the future, the prediction length type is considered to be long-term prediction, and the meteorological prediction data in the future 7-14 days can be collected at an interval of 24 hours, and each time point is every 24 hours.

[0032] In this way, based on different prediction length types, different data collection frequencies are adopted, which can realize accurate adaptation of data demand and prediction target, reduce the cost of data acquisition and processing, and also reduce the interference of data noise on the prediction result, and improve the prediction accuracy of the photovoltaic power generation amount.

[0033] In the embodiment of the present application, after obtaining the weather prediction data of the plurality of weather data sources, the weather prediction data can be preprocessed, for example, invalid data is filtered, the data format and data sequence of the weather prediction data of different weather data sources are unified, the time stamps of the weather prediction data of different weather data sources are aligned, and the weather prediction data array of each time point is generated, such as DateTime: {[data of the first weather data source], [data of the second weather data source], [data of the third weather data source]}.

[0034] In step 102, since the prediction accuracy of different weather data sources in different prediction scenarios can have a large difference, the target weight of each weather data source can be determined according to the first weight of each weather data source corresponding to the target weather type of each time point and the second weight of each weather data source corresponding to the prediction duration type of the to-be-predicted time range.

[0035] In specific implementation, the first weight is determined according to the weather type.

[0036] In one embodiment, before step 102, the first weight of each weather data source corresponding to the target weather type can also be determined according to the pre-set weight information of each weather data source corresponding to different weather types.

[0037] In specific implementation, the weight of each weather data source corresponding to different weather types can be set based on historical prediction data, for example, in sunny weather, local data is more important, since the third weather data source is strong in real-time and the local data is accurate, the weight of the first weather data source is set to 1, the weight of the second weather data source is set to 0.9, and the weight of the third weather data source is set to 0.8; in cloudy weather, regional data is more important, since the local prediction accuracy of the second weather data source is high, the weight of the first weather data source is set to 0.8, the weight of the second weather data source is set to 1, and the weight of the third weather data source is set to 0.9; in rainy weather, the global prediction model is more reliable, since the prediction position of the first weather data source covers a wide range, the weight of the first weather data source is set to 1, the weight of the second weather data source is set to 0.9, and the weight of the third weather data source is set to 0.7, and the like.

[0038] In this way, in specific application, the weight of each weather data source corresponding to the target weather type, i.e., the first weight, can be obtained from the pre-set weight information of each weather data source corresponding to different weather types according to the target weather type of each time point.

[0039] In specific implementation, the second weight is determined according to the prediction duration type.

[0040] In one embodiment, before step 102, the second weight of each weather data source can be determined through the following steps: determine a prediction length type according to length information of the time range to be predicted; obtain historical weather prediction data of the same prediction length type and corresponding historical real weather data of each weather data source in a preset historical time period; determine the second weight of each weather data source according to the deviation of the historical weather prediction data and the corresponding historical real weather data of each weather data source.

[0041] In a specific implementation, the second weight can be determined based on historical prediction accuracy information of the same prediction length type as the time range to be predicted. Specifically, first, the prediction length type to which the length of the time range to be predicted belongs is determined according to the length of the time range to be predicted. For example, the prediction length type can be short-term prediction (within 24 hours), medium-term prediction (1-6 days), or long-term prediction (7-14 days). Then, a historical time period, such as the past year, is set. Historical weather prediction data of the same prediction length type predicted by each weather data source in the historical time period and corresponding historical real weather data are obtained. For example, if it is short-term prediction, historical weather prediction data of each weather data source in which the prediction time range is not more than 24 hours in the past year and corresponding historical real weather data are obtained. Finally, the deviation of the historical weather prediction data and the corresponding historical real weather data of each weather data source is calculated, and the second weight of each weather data source is determined according to the deviation.

[0042] In one embodiment, the second weight of each weather data source is determined according to the deviation of the historical weather prediction data and the corresponding historical real weather data of each weather data source, and specifically can include: For each weather data source, error data of historical weather prediction data of different data types and corresponding historical real weather data at the same time point are calculated. The error data of different data types at the same time point are fused according to a preset third weight corresponding to different data types, to obtain comprehensive error of each time point. The second weight of each weather data source is determined according to the comprehensive error of each time point of each weather data source.

[0043] In implementation, the error between the historical meteorological prediction data of the same data type at the same time point and the corresponding historical real meteorological data is calculated; then, the errors of different data types at the same time point are fused according to the third weight corresponding to different data types, to obtain the comprehensive error of each time point, for example, the comprehensive error at a certain time point is: comprehensive error = temperature error × 0.4 + solar radiation intensity error × 0.6, wherein the temperature error is the data error of the data type of temperature, 0.4 is the third weight corresponding to the temperature, the solar radiation intensity error is the data error of the data type of solar radiation intensity, and 0.6 is the third weight corresponding to the solar radiation intensity; after obtaining the comprehensive error of each time point, the average value of the comprehensive error of each meteorological data source can be normalized to obtain the second weight of each meteorological data source.

[0044] After obtaining the first weight and the second weight of each meteorological data source, in step 102, the first weight and the second weight of each meteorological data source can be multiplied and then normalized to obtain the target weight of each meteorological data source.

[0045] In this way, based on the meteorological prediction characteristics of different meteorological types, different prediction length types, different meteorological data sources, and the importance of each data type, etc., the weights of each meteorological data source are determined, which can improve the reliability and data quality of the data of each meteorological data source, provide high-quality data basis for the prediction of photovoltaic power generation, and further improve the prediction accuracy of photovoltaic power generation.

[0046] In step 103, the meteorological prediction data of each time point in the to-be-predicted time range of the plurality of meteorological data sources can be weighted and fused according to the target weight of each meteorological data source to obtain the meteorological fused data of each time point.

[0047] In step 104, the meteorological fused data of each time point is input into the pre-trained photovoltaic power generation prediction model to obtain the predicted power generation information of each time point in the to-be-predicted time range.

[0048] The photovoltaic power generation prediction model can be obtained by training a machine learning model using the historical meteorological prediction data of the plurality of meteorological data sources and the corresponding historical photovoltaic power generation information.

[0049] For example, the photovoltaic power generation prediction model can be a model based on a long short-term memory (LSTM) neural network, or a model based on a gradient boosting decision tree (GBDT), or a model based on a support vector regression (SVR), or any combination of the above models.

[0050] In particular implementation, the historical meteorological data and the corresponding historical photovoltaic power generation information can be used as sample data to train the machine learning model to obtain the photovoltaic power generation prediction model.

[0051] In one embodiment, after step 103 and before step 104, the method further includes: The meteorological fusion data at each time point is subjected to feature extraction to obtain meteorological feature data at each time point, wherein the meteorological feature data includes temperature features, cloud coverage features, solar radiation intensity features, meteorological type features, hourly period features, daily period features, and / or seasonal period features. Step 104 specifically can include: The meteorological feature data at each time point is input into the pre-trained photovoltaic power generation prediction model to obtain predicted power generation information at each time point within the to-be-predicted time range.

[0052] In particular implementation, in order to improve the prediction accuracy of photovoltaic power generation, the present embodiment can extract multi-dimensional features in the meteorological fusion data, i.e., temperature features, which refer to ambient temperature data; cloud coverage features, which refer to cloud coverage percentage and affect the intensity of solar radiation reaching the ground; solar radiation intensity features, which directly affect photovoltaic power generation; meteorological type features, which refer to converting weather types such as sunny, cloudy, rainy, etc. into numerical codes; hourly period features, which refer to converting hourly time in a day into periodic numerical features; daily period features, which refer to converting dates in a year into periodic numerical features; and seasonal period features, which refer to the periodic features of the season corresponding to the date.

[0053] The above meteorological feature data can be input into the photovoltaic power generation prediction model to predict the photovoltaic power generation, so that through the above feature extraction, the photovoltaic power generation model can better capture various complex factors affecting power generation and their interactions, thereby improving the accuracy of photovoltaic power generation prediction.

[0054] Figure 2 Another flowchart of a multi-source data fusion photovoltaic power generation prediction method provided by the present embodiment is shown in FIG. 6. Figure 2 As shown in FIG. 6, the method includes the following steps: At step 201, the solar position information and the photovoltaic panel temperature information in a time range to be predicted are acquired. At step 104, the step can specifically include: The meteorological fusion data at each time point, the solar position information and the photovoltaic panel temperature information in the time range to be predicted are input into a pre-trained photovoltaic power generation prediction model to obtain the predicted power generation information at each time point in the time range to be predicted, and the photovoltaic power generation prediction model is obtained by training a machine learning model by using historical solar position information, historical photovoltaic panel temperature information, historical meteorological prediction data of multiple meteorological data sources, and corresponding historical photovoltaic power generation information.

[0055] In a specific implementation, in order to further improve the prediction accuracy, the solar position information and the photovoltaic panel temperature information are introduced to predict the photovoltaic power generation.

[0056] In one embodiment, the solar position information can include a solar altitude angle, which is an angle between the sun and the horizon and can directly affect the light intensity, and a solar azimuth angle, which is an angle of the sun relative to the north direction and affects the projection angle of light on the photovoltaic panel. The solar position information can be calculated based on the geographic position (latitude and longitude) and the prediction time information by using an existing astronomical algorithm.

[0057] In one embodiment, the photovoltaic panel temperature has a significant impact on the power generation efficiency of the photovoltaic system, and the efficiency decreases by about 0.4-0.5% for each 1°C increase in the photovoltaic panel temperature, so the photovoltaic panel temperature can be estimated based on the NOCT model. The NOCT (Nominal Operating Cell Temperature) is the cell operating temperature measured under standard test conditions (environmental temperature 20°C, solar radiation intensity 800W / m², wind speed 1m / s, and no load condition), and the NOCT is usually about 45°C. The photovoltaic panel temperature estimation formula based on the NOCT is: photovoltaic panel temperature = environmental temperature + (NOCT - 20°C) × (actual solar radiation intensity / 800W / m²).

[0058] In the embodiment of the present application, the solar position information and the photovoltaic panel temperature information can also be feature extracted to obtain a solar altitude angle feature, a solar azimuth angle feature, and a photovoltaic panel temperature feature.

[0059] In this way, Figure 2 The multi-source data fusion photovoltaic power generation prediction method shown in the figure introduces the solar position information and the photovoltaic panel temperature information to predict the photovoltaic power generation, can combine the multi-dimensional data such as the meteorological prediction data, the solar position information, and the photovoltaic panel temperature, and further improves the prediction accuracy of the photovoltaic power generation.

[0060] In summary, the photovoltaic power generation prediction method based on multi-source data fusion provided in this invention has the following beneficial effects: 1. By using meteorological forecast data from multiple meteorological data sources to predict photovoltaic power generation, the limitations of a single meteorological data source can be overcome, and the completeness of the data can be improved.

[0061] 2. When fusing meteorological forecast data from multiple meteorological data sources, the weights of the multiple meteorological data sources are dynamically set according to different meteorological types and forecast duration types for the time range to be predicted. This ensures the scientific and accurate allocation of weights, truly reflects the actual capabilities of each meteorological data source under different forecast scenarios, improves the reliability of meteorological forecast data, and thus improves the forecast accuracy of photovoltaic power generation.

[0062] 3. Solar position information and photovoltaic panel temperature information are incorporated to predict photovoltaic power generation, further improving the accuracy of photovoltaic power generation prediction.

[0063] This invention also provides a photovoltaic power generation prediction device based on multi-source data fusion, as described in the following embodiments. Since the principle by which this device solves the problem is similar to the aforementioned photovoltaic power generation prediction method based on multi-source data fusion, the implementation of this method can refer to the implementation of the photovoltaic power generation prediction method based on multi-source data fusion; repeated details will not be elaborated further.

[0064] like Figure 3 The diagram shown is a schematic of a photovoltaic power generation prediction device based on multi-source data fusion according to an embodiment of the present invention. The device may include the following modules: The first data acquisition module 301 is used to acquire meteorological forecast data from multiple meteorological data sources at various time points within the time range to be predicted, and to determine the target meteorological type at each time point. The weight determination module 302 is used to determine the target weight of each meteorological data source based on the first weight of each meteorological data source corresponding to the target meteorological type and the second weight of each meteorological data source corresponding to the prediction duration type of the time range to be predicted. The data fusion module 303 is used to fuse meteorological forecast data from multiple meteorological data sources at various time points within the time range to be predicted, based on the target weights of each meteorological data source, to obtain fused meteorological data at each time point. The model prediction module 304 is used to input the meteorological fusion data of each time point into the pre-trained photovoltaic power generation prediction model to obtain the predicted power generation information of each time point within the time range to be predicted. The photovoltaic power generation prediction model is obtained by training a machine learning model using historical meteorological prediction data from multiple meteorological data sources and their corresponding historical photovoltaic power generation information.

[0065] In an embodiment, the weather weight setting module can also be included for determining the first weight of each weather data source corresponding to the target weather type before the weight determining module 302 determines the target weight of each weather data source according to the first weight of each weather data source corresponding to the target weather type and the second weight of each weather data source corresponding to the prediction length type of the to-be-predicted time range: determining the first weight of each weather data source corresponding to the target weather type according to the pre-set weight information of each weather data source corresponding to different weather types.

[0066] In an embodiment, the length weight setting module can also be included for determining the prediction length type before the weight determining module 302 determines the target weight of each weather data source according to the first weight of each weather data source corresponding to the target weather type and the second weight of each weather data source corresponding to the prediction length type of the to-be-predicted time range: determining the prediction length type according to the length information of the to-be-predicted time range; obtaining historical weather prediction data of each weather data source for the same prediction length type and corresponding historical real weather data in a preset historical time period; determining the second weight of each weather data source according to the deviation of the historical weather prediction data of each weather data source and the corresponding historical real weather data.

[0067] In an embodiment, the length weight setting module can also be used for: for each weather data source: calculating error data of historical weather prediction data of different data types and corresponding historical real weather data at the same time point; and fusing the error data of different data types at the same time point according to a pre-set third weight corresponding to different data types to obtain comprehensive error at each time point; determining the second weight of each weather data source according to the comprehensive error of each time point of each weather data source.

[0068] In an embodiment, the feature extraction module can also be included for obtaining weather feature data at each time point by extracting features from the weather fusion data at each time point before the model prediction module 304 inputs the weather fusion data at each time point into the pre-trained photovoltaic power generation prediction model to obtain prediction power generation information at each time point in the to-be-predicted time range: extracting features from the weather fusion data at each time point to obtain weather feature data at each time point, wherein the weather feature data includes temperature features, cloud coverage features, solar radiation intensity features, weather type features, hourly period features, daily period features, and / or seasonal period features; The model prediction module 304 can be specifically used for: The meteorological feature data at each time point is input into a pre-trained photovoltaic power generation prediction model to obtain predicted power generation information at each time point in the to-be-predicted time range.

[0069] In one embodiment, the first data acquisition module 301 can be specifically used for: According to the prediction length type of the to-be-predicted time range, the data acquisition frequency is determined. According to the data acquisition frequency, meteorological prediction data in the to-be-predicted time range is collected from the multiple meteorological data sources to obtain meteorological prediction data at each time point of the multiple meteorological data sources.

[0070] In one embodiment, as shown in the figure, Figure 4 The device can further include a second data acquisition module 401: The solar position information and the photovoltaic panel temperature information in the to-be-predicted time range are acquired. The model prediction module 304 can be further used for: The meteorological fusion data at each time point, the solar position information and the photovoltaic panel temperature information in the to-be-predicted time range are input into a pre-trained photovoltaic power generation prediction model to obtain predicted power generation information at each time point in the to-be-predicted time range, and the photovoltaic power generation prediction model is obtained by training a machine learning model using historical solar position information, historical photovoltaic panel temperature information, historical meteorological prediction data of the multiple meteorological data sources, and corresponding historical photovoltaic power generation information.

[0071] The embodiment of the present application further provides a computer device, Figure 5 The computer device 500 includes a memory 510, a processor 520, and a computer program 530 stored in the memory 510 and executable on the processor 520, and the processor 520 implements the above-mentioned multi-source data fusion photovoltaic power generation prediction method when executing the computer program 530.

[0072] The embodiment of the present application further provides a computer readable storage medium, which stores a computer program, and the computer program is executed by a processor to implement the above-mentioned multi-source data fusion photovoltaic power generation prediction method.

[0073] The embodiment of the present application further provides a computer program product, which includes a computer program, and the computer program is executed by a processor to implement the above-mentioned multi-source data fusion photovoltaic power generation prediction method.

[0074] In the embodiment of the present application, the meteorological prediction data of each time point in the to-be-predicted time range of the plurality of meteorological data sources is acquired, and the target meteorological type of each time point is determined; the target weight of each meteorological data source is determined according to the first weight of each meteorological data source corresponding to the target meteorological type and the second weight of each meteorological data source corresponding to the prediction length type of the to-be-predicted time range; the meteorological prediction data of each time point in the to-be-predicted time range of the plurality of meteorological data sources is fused according to the target weight of each meteorological data source, to obtain the meteorological fused data of each time point; and the meteorological fused data of each time point is input into a pre-trained photovoltaic power generation prediction model to obtain the predicted power generation information of each time point in the to-be-predicted time range, the photovoltaic power generation prediction model being obtained by training a machine learning model by using historical meteorological data and corresponding historical photovoltaic power generation information. Compared with the existing method of predicting photovoltaic power generation by relying on a single meteorological data source, the embodiment of the present application predicts the photovoltaic power generation by fusing the meteorological prediction data of a plurality of meteorological data sources, and dynamically sets the target weight of each meteorological data source in combination with the meteorological type and the prediction length type of the to-be-predicted time range, thereby ensuring the scientificity and accuracy of the weight distribution, truly reflecting the actual capability of each meteorological data source in different prediction scenarios, and further improving the prediction accuracy of photovoltaic power generation.

[0075] Those skilled in the art will understand that the embodiments of the present application can be provided as methods, systems, or computer program products. Therefore, the present application can take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can take the form of a computer program product implemented on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROMs, optical storage, etc.) containing computer-usable program code.

[0076] The present application is described with reference to flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, and the combination of flows and / or blocks in the flowcharts and / or block diagrams can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing apparatus to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing apparatus produce a device that implements the functions specified in the flowcharts and / or block diagrams. Figure 1 The functions specified in a flow or multiple flows and / or blocks Figure 1 The device that implements the functions specified in a flow or multiple flows and / or blocks.

[0077] These computer program instructions can also be stored in a computer readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer readable memory produce an article of manufacture including instructions which implement the flow Figure 1 of the flow or flows and / or blocks Figure 1 of the block or blocks specified in the flow.

[0078] The computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the flow Figure 1 of the flow or flows and / or blocks Figure 1 of the block or blocks specified in the flow.

[0079] The specific embodiments described above are provided for purposes of illustration only and are not intended to limit the scope of the present application, which is defined by the following claims.

Claims

1. A method for photovoltaic power generation capacity prediction based on multi-source data fusion, characterized in that, The method comprises the following steps: acquiring meteorological prediction data of each time point in a to-be-predicted time range from a plurality of meteorological data sources, and determining a target meteorological type of each time point; determining a target weight of each meteorological data source according to a first weight of each meteorological data source corresponding to the target meteorological type and a second weight of each meteorological data source corresponding to a prediction length type of the to-be-predicted time range; fusing the meteorological prediction data of each time point in the to-be-predicted time range from the plurality of meteorological data sources according to the target weight of each meteorological data source, to obtain meteorological fused data of each time point; inputting the meteorological fused data of each time point into a pre-trained photovoltaic power generation prediction model to obtain prediction power generation information of each time point in the to-be-predicted time range, wherein the photovoltaic power generation prediction model is obtained by training a machine learning model by using historical meteorological data of the plurality of meteorological data sources and corresponding historical photovoltaic power generation information.

2. The multi-source data fused photovoltaic power generation amount prediction method of claim 1, wherein, Before the step of determining the target weight of each meteorological data source according to the first weight of each meteorological data source corresponding to the target meteorological type and the second weight of each meteorological data source corresponding to the prediction length type of the to-be-predicted time range, the method further comprises the following steps: determining the first weight of each meteorological data source corresponding to the target meteorological type according to pre-set meteorological data source weight information corresponding to different meteorological types.

3. The multi-source data fused photovoltaic power generation amount prediction method of claim 1, wherein, Before the step of determining the target weight of each meteorological data source according to the first weight of each meteorological data source corresponding to the target meteorological type and the second weight of each meteorological data source corresponding to the prediction length type of the to-be-predicted time range, the method further comprises the following steps: determining the prediction length type according to length information of the to-be-predicted time range; acquiring historical meteorological prediction data of each meteorological data source corresponding to the same prediction length type and corresponding historical real meteorological data in a preset historical time period; determining the second weight of each meteorological data source according to a deviation of the historical meteorological prediction data of each meteorological data source and the corresponding historical real meteorological data.

4. The multi-source data fused photovoltaic power generation amount prediction method of claim 3, wherein, The step of determining the second weight of each meteorological data source according to the deviation of the historical meteorological prediction data of each meteorological data source and the corresponding historical real meteorological data comprises the following steps: for each meteorological data source, calculating error data of different data types of historical meteorological prediction data and corresponding historical real meteorological data at the same time point; fusing the error data of different data types at the same time point according to pre-set third weights corresponding to different data types, to obtain comprehensive error of each time point; determining the second weight of each meteorological data source according to the comprehensive error of each time point of each meteorological data source.

5. The method for multi-source data fusion photovoltaic power generation prediction of claim 1, wherein, Before the step of inputting the meteorological fused data of each time point into the pre-trained photovoltaic power generation prediction model to obtain the prediction power generation information of each time point in the to-be-predicted time range, the method further comprises the following steps: performing feature extraction on the meteorological fused data of each time point to obtain meteorological feature data of each time point, wherein the meteorological feature data comprises temperature features, cloud coverage features, solar radiation intensity features, meteorological type features, hour-periodic features, date-periodic features, and / or seasonal periodic features. The meteorological fusion data of each time point is input into a pre-trained photovoltaic power generation prediction model to obtain predicted power generation information of each time point in the to-be-predicted time range. The meteorological fusion data of each time point is input into a pre-trained photovoltaic power generation prediction model to obtain predicted power generation information of each time point in the to-be-predicted time range.

6. The method for multi-source data fusion based photovoltaic power generation prediction of claim 1, wherein, The meteorological prediction data of each time point in the to-be-predicted time range of the plurality of meteorological data sources is obtained, including: According to the prediction length type of the to-be-predicted time range, the data acquisition frequency is determined; According to the data acquisition frequency, the meteorological prediction data in the to-be-predicted time range is collected from the plurality of meteorological data sources to obtain the meteorological prediction data of each time point of the plurality of meteorological data sources.

7. The multi-source data fused photovoltaic power generation amount prediction method according to any one of claims 1-6, characterized in that, Further comprising: Obtaining solar position information and photovoltaic panel temperature information in the to-be-predicted time range; The meteorological fusion data of each time point is input into a pre-trained photovoltaic power generation prediction model to obtain predicted power generation information of each time point in the to-be-predicted time range. The meteorological fusion data of each time point, the solar position information and the photovoltaic panel temperature information in the to-be-predicted time range are input into a pre-trained photovoltaic power generation prediction model to obtain predicted power generation information of each time point in the to-be-predicted time range, and the photovoltaic power generation prediction model is obtained by training a machine learning model using historical solar position information, historical photovoltaic panel temperature information, historical meteorological prediction data of the plurality of meteorological data sources, and corresponding historical photovoltaic power generation information.

8. A multi-source data fusion photovoltaic power generation amount prediction device, characterized by, Comprising: A first data acquisition module is configured to acquire meteorological prediction data of each time point in a to-be-predicted time range of a plurality of meteorological data sources, and determine a target meteorological type of each time point; A weight determination module is configured to determine a target weight of each meteorological data source according to a first weight of each meteorological data source corresponding to the target meteorological type and a second weight of each meteorological data source corresponding to a prediction length type of the to-be-predicted time range; A data fusion module is configured to fuse the meteorological prediction data of each time point in the to-be-predicted time range of the plurality of meteorological data sources according to the target weight of each meteorological data source to obtain meteorological fusion data of each time point. A model prediction module is configured to input the meteorological fusion data of each time point into a pre-trained photovoltaic power generation prediction model to obtain predicted power generation information of each time point in the to-be-predicted time range, and the photovoltaic power generation prediction model is obtained by training a machine learning model using historical meteorological prediction data of the plurality of meteorological data sources and corresponding historical photovoltaic power generation information.

9. A computer device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, The processor executes the computer program to implement the multi-source data fusion photovoltaic power generation prediction method of any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer readable storage medium stores a computer program, and the computer program is executed by the processor to implement the multi-source data fusion photovoltaic power generation prediction method of any one of claims 1 to 7.

11. A computer program product, characterised in that, The computer program product comprises a computer program, and the computer program is executed by the processor to implement the multi-source data fusion photovoltaic power generation prediction method of any one of claims 1 to 7.

Images