A photovoltaic power generation prediction method and device

Abstract

The application discloses a photovoltaic power generation prediction method and device, relates to the intelligent power generation technical field, solves the problems that data errors exist in data collection and power generation processes, and directly uses output results as prediction values to cause certain deviation between the prediction values and actual values; the application obtains historical power generation amounts of a plurality of photovoltaic panels, real-time weather data, historical weather data and corresponding historical total power generation amounts; divides a photovoltaic power generation area into a plurality of power generation areas based on the historical power generation amounts of the plurality of photovoltaic panels; analyzes power generation thresholds based on the historical weather data and the corresponding historical total power generation amounts; predicts power generation amounts of the plurality of areas based on the real-time weather data and the power generation thresholds, and obtains predicted power generation amounts; integrates the predicted power generation amounts of the plurality of areas to obtain photovoltaic predicted power generation amounts; and accurately predict the power generation amounts and reduce prediction errors caused by uncertain factors.

Description

Technical Field

[0001] This application belongs to the field of smart power generation and relates to photovoltaic power generation prediction technology, specifically a photovoltaic power generation prediction method and device. Background Technology

[0002] Photovoltaic power generation is a clean energy technology that uses the photoelectric effect of semiconductor materials to directly convert solar energy into electrical energy. Photovoltaic power generation is affected by factors such as sunlight intensity, weather, and season, resulting in large fluctuations in output power. By predicting the output of photovoltaic power generation, the output of other power sources can be reasonably adjusted to avoid grid frequency fluctuations or voltage instability caused by sudden drops or increases in photovoltaic output. By improving the prediction accuracy and consumption ratio of photovoltaic power generation, the dependence on traditional energy sources such as coal and gas power can be reduced, and the emissions of pollutants such as carbon dioxide and sulfur oxides can be reduced.

[0003] Existing photovoltaic (PV) power generation forecasting methods typically involve inputting weather data into a pre-trained artificial intelligence (AI) model to predict PV power generation. However, the environmental impact varies depending on the location of the PV panels, leading to low accuracy when predicting the entire PV power generation system as a whole. Furthermore, current technologies directly use the model's output as the predicted value, which introduces data errors during data acquisition and power generation. Directly using the output as the predicted value results in a discrepancy between the predicted and actual values.

[0004] This application provides a photovoltaic power generation prediction method and apparatus to solve the above-mentioned technical problems. Summary of the Invention

[0005] This application aims to at least solve one of the technical problems existing in the prior art; to this end, this application proposes a photovoltaic power generation prediction method and device to solve the problem that the environmental impact varies depending on the location of the photovoltaic panel, and predicting the entire photovoltaic power generation as a whole will lead to low prediction accuracy; at the same time, the prior art directly uses the output of the model as the predicted value. However, there are data errors in the data acquisition and power generation process, and directly using the output as the predicted value will result in a certain deviation between the predicted value and the actual value.

[0006] To achieve the above objectives, the first aspect of this application provides a photovoltaic power generation prediction method, comprising: Obtain historical power generation data, real-time weather data, historical weather data, and corresponding historical total power generation data for several photovoltaic panels; The photovoltaic power generation area is divided into several power generation areas based on the historical power generation of several photovoltaic panels. Analyze power generation thresholds based on historical weather data and corresponding historical total power generation. Based on real-time weather data and power generation thresholds, the power generation of several regions is predicted to obtain the predicted power generation. The predicted power generation of several regions is integrated to obtain the predicted photovoltaic power generation.

[0007] Based on the above, this application obtains multi-dimensional data such as historical power generation of photovoltaic panels, real-time weather data, historical weather data, and corresponding historical total power generation, enabling a more comprehensive consideration of various factors affecting photovoltaic power generation. It divides the photovoltaic power generation area into several power generation zones and conducts independent analysis and prediction based on the historical power generation data of each zone. Through zone division, more accurate modeling and prediction can be performed for the unique characteristics of each zone, avoiding the neglect of local differences while treating the entire power generation area as a whole, further improving the accuracy of prediction. By analyzing power generation thresholds using historical weather data and corresponding historical total power generation, a reasonable range of photovoltaic power generation under different weather conditions can be identified. This helps determine whether the power generation of each zone is at a normal level under real-time weather conditions, thereby more accurately predicting power generation and reducing prediction errors caused by uncertainties such as weather changes. Predicting photovoltaic power generation allows for better coordination of energy production and consumption, improves the grid's capacity to absorb renewable energy, and promotes the development of smart grids towards greater efficiency, reliability, and greenness.

[0008] Preferably, the division of the photovoltaic power generation area into several power generation areas based on the historical power generation of several photovoltaic panels includes: Retrieve the historical power generation data of several photovoltaic panels; integrate the historical power generation data of several photovoltaic panels according to the daily power generation data to obtain the historical daily power generation data; align the historical daily power generation data of several photovoltaic panels by time. Retrieve the historical daily power generation of several photovoltaic panels within a preset time period; calculate the average daily power generation of several photovoltaic panels; calculate the difference in average daily power generation between adjacent photovoltaic panels; When the difference is less than the preset difference threshold, the corresponding photovoltaic panels are divided into the same area, and the average daily power generation of the corresponding area is recalculated; the difference calculation and division steps are repeated to obtain several power generation areas.

[0009] Preferably, the step of analyzing the power generation threshold based on historical weather data and corresponding historical total power generation includes: Retrieve historical weather data and corresponding historical total power generation; the historical weather data includes: solar radiation, historical ambient temperature, historical relative humidity, historical wind speed and direction; The 3σ principle is used to identify outliers in historical weather data, and interpolation is used to replace outliers in historical weather data; in addition, the historical weather data is standardized. Historical wind direction is converted into sine and cosine components; the processed historical weather data is then labeled as weather feature vectors. The critical points are identified based on weather feature vectors to obtain the corresponding power generation thresholds. in, This represents the i-th solar radiation; This represents the i-th historical ambient temperature; This represents the i-th historical relative humidity. Indicates the i-th historical wind speed; Represents the sine component of the i-th historical wind direction; Let represent the cosine component of the i-th historical wind direction.

[0010] Preferably, the step of identifying the critical point based on the weather feature vector to obtain the corresponding power generation threshold includes: Retrieve several weather feature vectors; integrate these weather feature vectors with their corresponding historical total power generation to obtain several samples; select samples with a historical power generation of 0 from these samples as zero samples; mark the remaining samples as positive samples. The distance between several positive samples and zero samples is calculated based on the boundary calculation function; the minimum value is selected as the final distance between the positive sample and the zero sample region; the top m positive samples with the smallest final distance are selected to form the boundary sample set. The expression for the boundary calculation function is: ;in, This represents the p-th positive sample; This represents the z-th zero sample; This represents the distance between the p-th positive sample and the z-th zero sample. This represents the k-th feature in the p-th positive sample; This represents the k-th feature in the z-th zero sample; The minimum value of each feature value in the boundary sample set is calculated using the norm minimum value analysis function; the expression of the norm minimum value analysis function is: ;in, This represents the k-th feature of the boundary sample; Represents positive samples in the boundary sample set; Through formula The minimum value of the boundary sample is calculated; the minimum value of the boundary sample is marked as the power generation threshold; where, Represents the boundary sample set; This represents the minimum value function.

[0011] Preferably, the step of predicting the power generation of several regions based on real-time weather data and power generation thresholds to obtain the predicted power generation includes: Retrieve real-time weather data and power generation thresholds; real-time weather data includes: solar radiation, real-time ambient temperature, relative humidity, real-time wind speed and direction; The real-time weather data is compared with the corresponding power generation threshold. If all the sub-data of the real-time weather data are greater than the corresponding power generation threshold, the power generation of several areas is predicted based on the real-time weather data. Otherwise, the predicted power generation is directly marked as 0.

[0012] Preferably, the step of predicting power generation in several regions based on real-time weather data includes: Retrieve real-time weather data for a specified time period; integrate the real-time weather data for the specified time period in chronological order to obtain a weather impact sequence; The power generation prediction model is invoked by inputting the weather impact sequence into the power generation prediction model to obtain the predicted power generation of several regions; the power generation prediction model is built based on an artificial intelligence model.

[0013] Preferably, the power prediction model is built based on an artificial intelligence model, including: Obtain the standard dataset; the standard dataset includes standard input data consistent with the content attributes of weather impact sequences, and standard output data consistent with the content attributes of power generation in several regions; Select model frameworks and deep learning algorithms from the artificial intelligence library; construct the model based on the deep learning algorithm and model framework to obtain the constructed model; The standard dataset is divided into training, validation, and test sets according to a preset ratio; the training set is used to train the model; the validation set is used to adjust the internal parameters of the model; and the test set is used to test the trained model and obtain test metrics. Obtain the indicator threshold; compare the test indicator with the indicator threshold; if all test indicators are greater than the indicator threshold, mark the constructed model as the power prediction model; otherwise, reconstruct and train the constructed model.

[0014] It should be noted that the test metrics include: accuracy, F1 score, recall, and stability; the thresholds for the metrics and the split ratio of the standard dataset are set by professional technicians according to the actual situation; when rebuilding the model, the model framework or deep learning algorithm can be changed; when retraining the model, the split ratio of the standard dataset can be adjusted.

[0015] Preferably, the process of integrating the predicted power generation from several regions to obtain the predicted photovoltaic power generation includes: Retrieve the predicted power generation for several regions; obtain the prediction adjustment function; The predicted power generation of several regions is adjusted based on the prediction adjustment function to obtain the final power generation; the photovoltaic predicted power generation is obtained by calculating the sum of the final power generation of several regions.

[0016] The second aspect of this application provides a photovoltaic power generation prediction device, including: a data acquisition module, a condition division module, and a power generation prediction module; The data acquisition module is used to acquire the historical power generation of several photovoltaic panels, real-time weather data, historical weather data, and the corresponding historical total power generation. The division condition module is used to divide the photovoltaic power generation area into several power generation areas based on the historical power generation of several photovoltaic panels; and to analyze the power generation threshold based on historical weather data and the corresponding historical total power generation. The power generation prediction module is used to predict the power generation of several regions based on real-time weather data and power generation thresholds to obtain the predicted power generation; and to integrate the predicted power generation of several regions to obtain the photovoltaic predicted power generation.

[0017] A third aspect of this application provides a computer-readable storage medium storing instructions for performing the methods described in the first aspect and any possible implementation thereof.

[0018] Compared with the prior art, the beneficial effects of this application are: 1. This application integrates the historical power generation of several photovoltaic panels into historical daily power generation and aligns the data by time, unifying the time dimension of the data. This helps eliminate data chaos caused by time inconsistencies, allowing subsequent analysis to be based on a neat and orderly dataset, thus improving the accuracy and efficiency of data processing. The application uses the 3σ principle to identify outliers in historical weather data and replaces them using interpolation, effectively removing noise and unreasonable values ​​from the data and preventing abnormal data from interfering with subsequent analysis. The application calculates the average daily power generation of several photovoltaic panels within a preset time period and the difference between adjacent values. Based on the comparison of the difference with a preset threshold, power generation areas are divided, and this process is repeated. The calculation and partitioning steps consider the similarity of power generation capacity of different photovoltaic panels, ensuring that photovoltaic panels within the same region have similar power generation characteristics. This helps to more accurately analyze the power generation patterns of each region and improve the accuracy of subsequent regional power generation predictions. Historical wind direction is converted into sine and cosine components and combined with other historical weather data to construct a weather feature vector, comprehensively considering various weather factors affecting photovoltaic power generation. The power generation threshold is obtained by identifying critical points based on the weather feature vector, enabling more accurate capture of the boundary of photovoltaic power generation changes under different weather conditions. The weather feature vector is integrated with historical power generation data into samples, which are then divided into zero samples and positive samples. The distance from the positive samples to the zero samples is calculated using a boundary calculation function to select the boundary sample set. The minimum value of each feature value in the boundary sample set is then calculated using a norm minimum value analysis function, ultimately determining the boundary samples as the power generation threshold. This approach allows for more accurate identification of power generation boundary conditions, making the power generation threshold more consistent with actual conditions and improving the reliability of power generation predictions.

[0019] 2. This application compares real-time weather data with power generation thresholds. Power generation prediction is only performed when all sub-data points exceed the corresponding power generation thresholds; otherwise, it is directly marked as 0. This avoids unnecessary predictions under adverse weather conditions and reduces prediction errors caused by unsuitable weather conditions. Real-time weather data for a set time period is integrated into a weather impact sequence and input into the model for prediction, fully considering the impact of the continuity of weather changes and time series characteristics on power generation. Compared to using weather data from a single moment for prediction, this method can more comprehensively capture weather change trends and improve prediction accuracy. The prediction adjustment function comprehensively considers factors such as temperature loss and error fluctuations, making comprehensive adjustments to the predicted power generation to obtain a more accurate final power generation. The sum of the final power generation of several regions is calculated to obtain the photovoltaic predicted power generation, providing a reliable basis for energy management and dispatch. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0021] Figure 1 This is a schematic diagram illustrating the overall steps of the method described in this application; Figure 2 This is a schematic diagram illustrating the steps involved in dividing the region and determining the threshold in this application. Figure 3 This is a schematic diagram illustrating the power generation forecasting and adjustment steps in this application; Figure 4 This is a schematic diagram of the structural connection of the device modules in this application. Detailed Implementation

[0022] The technical solutions of this application will be clearly and completely described below with reference to the embodiments. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.

[0023] Please see Figure 1 The first aspect of this application provides a photovoltaic power generation prediction method and apparatus, including: S101. Obtain the historical power generation of several photovoltaic panels, real-time weather data, historical weather data, and the corresponding historical total power generation. S102. Divide the photovoltaic power generation area into several power generation areas based on the historical power generation of several photovoltaic panels; S103. Analyze power generation thresholds based on historical weather data and corresponding historical total power generation. S104. Based on real-time weather data and power generation thresholds, predict the power generation of several regions to obtain the predicted power generation. S105. Integrate the predicted power generation of several regions to obtain the predicted photovoltaic power generation.

[0024] This application acquires multi-dimensional data, including historical power generation of photovoltaic panels, real-time weather data, historical weather data, and corresponding historical total power generation, enabling a more comprehensive consideration of various factors affecting photovoltaic power generation. It divides the photovoltaic power generation area into several regions and conducts independent analysis and prediction based on historical power generation data for each region. This regional division allows for more accurate modeling and prediction tailored to the unique characteristics of each region, avoiding the neglect of local differences by treating the entire power generation area as a whole, thus further improving prediction accuracy. Utilizing historical weather data and corresponding historical total power generation to analyze power generation thresholds allows for identifying reasonable ranges for photovoltaic power generation under different weather conditions. This helps determine whether the power generation in each region is at a normal level under real-time weather conditions, thereby more accurately predicting power generation and reducing prediction errors caused by uncertainties such as weather changes. Predicting photovoltaic power generation allows for better coordination of energy production and consumption, improves the grid's capacity to absorb renewable energy, and promotes the development of smart grids towards greater efficiency, reliability, and greenness.

[0025] In one possible implementation of the embodiments of this application, combined with Figure 1 ,like Figure 2 As shown, the above S102-S103 can be specifically implemented through the following S201-S208, which are explained in detail below: S201. Retrieve the historical power generation of several photovoltaic panels; integrate the historical power generation of several photovoltaic panels according to the daily power generation to obtain the historical daily power generation; align the historical daily power generation of several photovoltaic panels by time.

[0026] S202. Retrieve the historical daily power generation of several photovoltaic panels within a preset time period; calculate the average daily power generation of several photovoltaic panels; calculate the difference in the average daily power generation of adjacent photovoltaic panels.

[0027] S203. When the difference is less than the preset difference threshold, the corresponding photovoltaic panels are divided into the same area, and the average daily power generation of the corresponding area is recalculated; and the difference calculation and division steps are repeated to obtain several power generation areas.

[0028] Example: Suppose a photovoltaic power station has 3 photovoltaic panels (A, B, C) and corresponding meteorological station data, with a time range of January 1, 2025 to January 7, 2025; Photovoltaic panel A generated 120 kWh on January 1st and 115 kWh on January 2nd; panel B generated 118 kWh on January 1st and 112 kWh on January 2nd; panel C generated 95 kWh on January 1st and 90 kWh on January 2nd. After aligning the dates, a daily power generation matrix is ​​formed; the regions are divided using January 1st as an example. Assume the preset difference threshold is 5 kWh. The difference between A and B on January 1st is |120-118| = 2 kWh < 5, so they are classified as Region 1; the difference between B and C is |118-95| = ​​23 kWh > 5, so C is classified as Region 2. The average power generation in Region 1 is recalculated as (120+118) / 2 = 119 kWh, and in Region 2 it is 95 kWh.

[0029] S204. Retrieve historical weather data and corresponding historical total power generation; identify outliers in historical weather data using the 3σ principle, replace outliers in historical weather data using interpolation; and standardize historical weather data.

[0030] Historical weather data includes: solar radiation, historical ambient temperature, historical relative humidity, and historical wind speed and direction.

[0031] S205. Convert historical wind direction into sine and cosine components; label the processed historical weather data as weather feature vectors. The critical points are identified based on weather feature vectors to obtain the corresponding power generation thresholds. in, This represents the i-th solar radiation; This represents the i-th historical ambient temperature; This represents the i-th historical relative humidity. Indicates the i-th historical wind speed; Represents the sine component of the i-th historical wind direction; Let represent the cosine component of the i-th historical wind direction.

[0032] Example: The raw solar radiation data for January 3rd is 1200 W / m² (historical average 800 W / m², standard deviation 150). The upper limit of 3σ is 800 + 3 × 150 = 1250, which is within the limit. The humidity data for January 4th is 95% (historical average 60%, standard deviation 10). The upper limit of 3σ is 60 + 3 × 10 = 90%. 95% > 90% is considered an outlier and is replaced with 75% using linear interpolation. On January 5th, the wind direction of 225° is converted to a sine component sin(225°)≈-0.707, a cosine component cos(225°)≈-0.707, and an eigenvector X_i=(900,25,60,5,-0.707,-0.707).

[0033] S206. Retrieve several weather feature vectors; integrate the several weather feature vectors with the corresponding historical total power generation to obtain several samples; select samples with a historical power generation of 0 from the several samples as zero samples; mark the remaining samples as positive samples. S207. Calculate the distance between several positive samples and zero samples based on the boundary calculation function; select the minimum value as the final distance between the positive sample and the zero sample region; select the top m positive samples with the smallest final distance to form the boundary sample set; use the norm minimum value analysis function to calculate the minimum value of each feature value in the boundary sample set.

[0034] The expression for the boundary calculation function is as follows: ; This represents the p-th positive sample; This represents the z-th zero sample; This represents the distance between the p-th positive sample and the z-th zero sample. This represents the k-th feature in the p-th positive sample; This represents the k-th feature in the z-th zero sample; The expression for the norm minimum analysis function is: ;in, This represents the k-th feature of the boundary sample; Represents positive samples in the boundary sample set; S208, through formula The minimum value of the boundary sample is calculated; the minimum value of the boundary sample is marked as the power generation threshold.

[0035] in, Represents the boundary sample set; This represents the minimum value function.

[0036] Example: January 6th power generation was 0 kWh (zero sample), other dates are positive samples. Calculate the Euclidean distance between the positive samples and the zero samples, such as the distance from the January 1st sample X_p=(900,25,60,5,-0.707,-0.707) to the zero sample. ; Select the two positive samples with the smallest distance (e.g., January 1st and January 2nd), and calculate the minimum features of the boundary sample set: minimum solar radiation 850W / m², minimum temperature 20℃, minimum humidity 55%, minimum wind speed 3m / s, minimum wind direction sine -0.5, minimum cosine -0.5; The minimum L2 norm of the boundary samples was calculated using the argmin function. The final power generation thresholds were set as follows: solar radiation threshold 850W / m², temperature threshold 20℃, humidity threshold 55%, wind speed threshold 3m / s, wind direction sine threshold -0.5, and wind direction cosine threshold -0.5.

[0037] Based on the above steps, this application integrates the historical power generation of several photovoltaic panels into historical daily power generation and aligns the data by time, unifying the time dimension of the data. This helps eliminate data chaos caused by time inconsistencies, allowing subsequent analysis to be based on a neat and orderly dataset, thus improving the accuracy and efficiency of data processing. The application uses the 3σ principle to identify outliers in historical weather data and replaces them using interpolation, effectively removing noise and unreasonable values ​​from the data and preventing abnormal data from interfering with subsequent analysis. The application calculates the average daily power generation of several photovoltaic panels within a preset time period and the difference between adjacent values, and divides the power generation areas based on the comparison of the difference with a preset threshold. The calculation and division steps are repeated; the similarity of power generation capacity of different photovoltaic panels is considered, so that photovoltaic panels in the same area have similar power generation characteristics, which helps to analyze the power generation patterns of each region more accurately and improve the accuracy of subsequent regional power generation prediction; historical wind direction is converted into sine and cosine components, and weather feature vectors are constructed with other historical weather data to comprehensively consider various weather factors affecting photovoltaic power generation; the critical point is identified based on the weather feature vector to obtain the power generation threshold, which can more accurately capture the boundary of photovoltaic power generation change under different weather conditions; the weather feature vector and historical power generation are integrated into samples and divided into zero samples and positive samples. The distance from the positive sample to the zero sample is calculated by the boundary calculation function, the boundary sample set is screened, and then the minimum value of each feature value in the boundary sample set is calculated by the norm minimum value analysis function, and finally the boundary sample is determined as the power generation threshold; this can more accurately find the boundary situation of power generation, making the power generation threshold more consistent with the actual situation and improving the reliability of power generation prediction.

[0038] In one possible implementation of the embodiments of this application, combined with Figure 1 ,like Figure 3 As shown, the above S104-S105 can be specifically implemented through the following S301-S303, which are explained in detail below: S301. Retrieve real-time weather data and power generation thresholds; compare the real-time weather data with the corresponding power generation thresholds; when all sub-data of the real-time weather data are greater than the corresponding power generation thresholds, predict the power generation of several areas based on the real-time weather data; otherwise, directly mark the predicted power generation as 0.

[0039] The real-time weather data includes: solar radiation, real-time ambient temperature, relative humidity, and real-time wind speed and direction.

[0040] S302. Retrieve real-time weather data for a set time period; integrate the real-time weather data for the set time period in chronological order to obtain a weather impact sequence; call the power generation prediction model, input the weather impact sequence into the power generation prediction model, and obtain the predicted power generation for several regions.

[0041] In one possible implementation, the power prediction model is built based on an artificial intelligence model, including: Obtain the standard dataset; the standard dataset includes standard input data consistent with the content attributes of weather impact sequences, and standard output data consistent with the content attributes of power generation in several regions; Select model frameworks and deep learning algorithms from the artificial intelligence library; construct the model based on the deep learning algorithm and model framework to obtain the constructed model; The standard dataset is divided into training, validation, and test sets according to a preset ratio; the training set is used to train the model; the validation set is used to adjust the internal parameters of the model; and the test set is used to test the trained model and obtain test metrics. Obtain the indicator threshold; compare the test indicator with the indicator threshold; if all test indicators are greater than the indicator threshold, mark the constructed model as the power prediction model; otherwise, reconstruct and train the constructed model.

[0042] It should be noted that the test metrics include: accuracy, F1 score, recall, and stability; the thresholds for the metrics and the split ratio of the standard dataset are set by professional technicians according to the actual situation; when rebuilding the model, the model framework or deep learning algorithm can be changed; when retraining the model, the split ratio of the standard dataset can be adjusted.

[0043] S303. Retrieve the predicted power generation of several regions; obtain the prediction adjustment function; adjust the predicted power generation of several regions based on the prediction adjustment function; obtain the final power generation; obtain the photovoltaic predicted power generation by calculating the sum of the final power generation of several regions.

[0044] Based on the above steps, this application compares real-time weather data with power generation thresholds. Power generation prediction is only performed when all sub-data are greater than the corresponding power generation thresholds; otherwise, it is directly marked as 0. This avoids unnecessary predictions under adverse weather conditions and reduces prediction errors caused by unsuitable weather conditions. Real-time weather data for a set time period is integrated into a weather impact sequence and input into the model for prediction, fully considering the impact of the continuity of weather changes and time series characteristics on power generation. Compared with using weather data at a single moment for prediction, this method can more comprehensively capture weather change trends and improve prediction accuracy. The prediction adjustment function comprehensively considers factors such as temperature loss and error fluctuations, making comprehensive adjustments to the predicted power generation to obtain a more accurate final power generation. The sum of the final power generation of several regions is calculated to obtain the photovoltaic predicted power generation, providing a reliable basis for energy management and dispatch.

[0045] Please see Figure 4The second aspect of this application provides a photovoltaic power generation prediction device, including: a data acquisition module, a division condition module, and a power generation prediction module; The data acquisition module is used to acquire the historical power generation of several photovoltaic panels, real-time weather data, historical weather data, and the corresponding historical total power generation. The division condition module is used to divide the photovoltaic power generation area into several power generation areas based on the historical power generation of several photovoltaic panels; and to analyze the power generation threshold based on historical weather data and the corresponding historical total power generation. The power generation prediction module is used to predict the power generation of several regions based on real-time weather data and power generation thresholds to obtain the predicted power generation; and to integrate the predicted power generation of several regions to obtain the photovoltaic predicted power generation.

[0046] A third aspect of this application provides a computer-readable storage medium storing instructions for performing the methods described in the first aspect and any possible implementation thereof.

[0047] Some of the data in the above formula are calculated by removing dimensions and taking their numerical values. The formula is the closest to the real situation obtained by software simulation of a large amount of collected data. The preset parameters and preset thresholds in the formula are set by those skilled in the art according to the actual situation or obtained through simulation of a large amount of data.

[0048] The working principle of this application is as follows: This application obtains the historical power generation of several photovoltaic panels, real-time weather data, historical weather data, and the corresponding historical total power generation; based on the historical power generation of several photovoltaic panels, the photovoltaic power generation area is divided into several power generation areas; based on the historical weather data and the corresponding historical total power generation, the power generation threshold is analyzed; based on the real-time weather data and the power generation threshold, the power generation of several areas is predicted to obtain the predicted power generation; the predicted power generation of several areas is integrated to obtain the photovoltaic predicted power generation.

[0049] The above embodiments are only used to illustrate the technical methods of this application and are not intended to limit it. Although this application has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical methods of this application without departing from the spirit and scope of the technical methods of this application.

Claims

1. A method for predicting photovoltaic power generation, characterized in that, include: Obtain historical power generation data, real-time weather data, historical weather data, and corresponding historical total power generation data for several photovoltaic panels; The photovoltaic power generation area is divided into several power generation areas based on the historical power generation of several photovoltaic panels. Analyze power generation thresholds based on historical weather data and corresponding historical total power generation. Based on real-time weather data and power generation thresholds, the power generation of several regions is predicted to obtain the predicted power generation. The predicted power generation of several regions is integrated to obtain the predicted photovoltaic power generation. The analysis of power generation thresholds based on historical weather data and corresponding historical total power generation includes: Retrieve historical weather data and corresponding historical total power generation; the historical weather data includes: solar radiation, historical ambient temperature, historical relative humidity, historical wind speed and direction; The 3σ principle is used to identify outliers in historical weather data, and interpolation is used to replace outliers in historical weather data; in addition, the historical weather data is standardized. Historical wind direction is converted into sine and cosine components; the processed historical weather data is then labeled as weather feature vectors. The critical points are identified based on weather feature vectors to obtain the corresponding power generation thresholds. in, This represents the i-th solar radiation; This represents the i-th historical ambient temperature; This represents the i-th historical relative humidity. Indicates the i-th historical wind speed; Represents the sine component of the i-th historical wind direction; Represents the cosine component of the i-th historical wind direction; The critical points are identified based on weather feature vectors to obtain the corresponding power generation thresholds, including: Retrieve several weather feature vectors; integrate these weather feature vectors with their corresponding historical total power generation to obtain several samples; select samples with a historical power generation of 0 from these samples as zero samples; mark the remaining samples as positive samples. The distance between several positive samples and zero samples is calculated based on the boundary calculation function; the minimum value is selected as the final distance between the positive sample and the zero sample region; the top m positive samples with the smallest final distance are selected to form the boundary sample set. The expression for the boundary calculation function is: ;in, This represents the p-th positive sample; This represents the z-th zero sample; This represents the distance between the p-th positive sample and the z-th zero sample. This represents the k-th feature in the p-th positive sample; This represents the k-th feature in the z-th zero sample; The minimum value of each feature value in the boundary sample set is calculated using the norm minimum value analysis function; the expression of the norm minimum value analysis function is: ;in, This represents the k-th feature of the boundary sample; Represents positive samples in the boundary sample set; Through formula The minimum value of the boundary sample is calculated; the minimum value of the boundary sample is marked as the power generation threshold; where, Represents the boundary sample set; This represents the minimum value function.

2. The photovoltaic power generation prediction method according to claim 1, characterized in that, The photovoltaic power generation area is divided into several power generation areas based on the historical power generation of several photovoltaic panels, including: Retrieve historical power generation data from several photovoltaic panels; integrate the historical power generation data of several photovoltaic panels according to daily power generation data to obtain historical daily power generation data; align the historical daily power generation data of several photovoltaic panels by time. Retrieve the historical daily power generation of several photovoltaic panels within a preset time period; calculate the average daily power generation of several photovoltaic panels; calculate the difference in average daily power generation between adjacent photovoltaic panels; When the difference is less than the preset difference threshold, the corresponding photovoltaic panels are divided into the same area, and the average daily power generation of the corresponding area is recalculated; the difference calculation and division steps are repeated to obtain several power generation areas.

3. The photovoltaic power generation prediction method according to claim 1, characterized in that, The method of predicting power generation in several regions based on real-time weather data and power generation thresholds to obtain predicted power generation includes: Retrieve real-time weather data and power generation thresholds; real-time weather data includes: solar radiation, real-time ambient temperature, relative humidity, real-time wind speed and direction; The real-time weather data is compared with the corresponding power generation threshold. If all the sub-data of the real-time weather data are greater than the corresponding power generation threshold, the power generation of several areas is predicted based on the real-time weather data. Otherwise, the predicted power generation is directly marked as 0.

4. The photovoltaic power generation prediction method according to claim 3, characterized in that, The prediction of power generation in several regions based on real-time weather data includes: Retrieve real-time weather data for a specified time period; integrate the real-time weather data for the specified time period in chronological order to obtain a weather impact sequence; The power generation prediction model is invoked by inputting the weather impact sequence into the power generation prediction model to obtain the predicted power generation of several regions; the power generation prediction model is built based on an artificial intelligence model.

5. The photovoltaic power generation prediction method according to claim 4, characterized in that, The power prediction model is built based on an artificial intelligence model and includes: Obtain the standard dataset; the standard dataset includes standard input data consistent with the content attributes of weather impact sequences, and standard output data consistent with the content attributes of power generation in several regions; Select model frameworks and deep learning algorithms from the artificial intelligence library; construct the model based on the deep learning algorithm and model framework to obtain the constructed model; The standard dataset is divided into training, validation, and test sets according to a preset ratio; the training set is used to train the model; the validation set is used to adjust the internal parameters of the model; and the test set is used to test the trained model and obtain test metrics. Obtain the indicator threshold; compare the test indicator with the indicator threshold; if all test indicators are greater than the indicator threshold, mark the constructed model as the power prediction model; otherwise, reconstruct and train the constructed model.

6. The photovoltaic power generation prediction method according to claim 1, characterized in that, The process of integrating the predicted power generation from several regions to obtain the predicted photovoltaic power generation includes: Retrieve the predicted power generation for several regions; obtain the prediction adjustment function; The predicted power generation of several regions is adjusted based on the prediction adjustment function to obtain the final power generation; the photovoltaic predicted power generation is obtained by calculating the sum of the final power generation of several regions.

7. A photovoltaic power generation prediction device, applied to the photovoltaic power generation prediction method according to any one of claims 1-6, characterized in that, include: The module includes a data acquisition module, a segmentation condition module, and a power prediction module. The data acquisition module is used to acquire the historical power generation of several photovoltaic panels, real-time weather data, historical weather data, and the corresponding historical total power generation. The division condition module is used to divide the photovoltaic power generation area into several power generation areas based on the historical power generation of several photovoltaic panels; and to analyze the power generation threshold based on historical weather data and the corresponding historical total power generation. The power generation prediction module is used to predict the power generation of several regions based on real-time weather data and power generation thresholds to obtain the predicted power generation; and to integrate the predicted power generation of several regions to obtain the photovoltaic predicted power generation. The analysis of power generation thresholds based on historical weather data and corresponding historical total power generation includes: Retrieve historical weather data and corresponding historical total power generation; the historical weather data includes: solar radiation, historical ambient temperature, historical relative humidity, historical wind speed and direction; The 3σ principle is used to identify outliers in historical weather data, and interpolation is used to replace outliers in historical weather data; in addition, the historical weather data is standardized. Historical wind direction is converted into sine and cosine components; the processed historical weather data is then labeled as weather feature vectors. The critical points are identified based on weather feature vectors to obtain the corresponding power generation thresholds. in, This represents the i-th solar radiation; This represents the i-th historical ambient temperature; This represents the i-th historical relative humidity. Indicates the i-th historical wind speed; Represents the sine component of the i-th historical wind direction; Represents the cosine component of the i-th historical wind direction; The critical points are identified based on weather feature vectors to obtain the corresponding power generation thresholds, including: Retrieve several weather feature vectors; integrate these weather feature vectors with their corresponding historical total power generation to obtain several samples; select samples with a historical power generation of 0 from these samples as zero samples; mark the remaining samples as positive samples. The distance between several positive samples and zero samples is calculated based on the boundary calculation function; the minimum value is selected as the final distance between the positive sample and the zero sample region; the top m positive samples with the smallest final distance are selected to form the boundary sample set. The expression for the boundary calculation function is: ;in, This represents the p-th positive sample; This represents the z-th zero sample; This represents the distance between the p-th positive sample and the z-th zero sample. This represents the k-th feature in the p-th positive sample; This represents the k-th feature in the z-th zero sample; The minimum value of each feature value in the boundary sample set is calculated using the norm minimum value analysis function; the expression of the norm minimum value analysis function is: ;in, This represents the k-th feature of the boundary sample; Represents positive samples in the boundary sample set; Through formula The minimum value of the boundary sample is calculated; the minimum value of the boundary sample is marked as the power generation threshold; where, Represents the boundary sample set; This represents the minimum value function.

Images