Distributed photovoltaic power prediction method, device and equipment and readable storage medium

Abstract

The invention relates to a distributed photovoltaic power prediction method, device and equipment and a readable storage medium. The method comprises the following steps: acquiring generated power data of each power station in a region; performing clustering processing on the plurality of power stations based on the generated power data, and determining the power station category to which each power station belongs; for each power station category, determining target irradiance data acquired by a target meteorological data acquisition site corresponding to the power station category; and predicting the generated power of the power station category based on the target irradiance data to obtain target predicted power. By adopting the method, the power prediction accuracy can be improved.

Description

Technical Field

[0001] The present application relates to the field of photovoltaic power generation technology, and in particular to a distributed photovoltaic power prediction method, device, equipment and readable storage medium. Background Art

[0002] With the development of new energy, distributed photovoltaic power generation has been widely used around the world as a clean, efficient and sustainable way of energy utilization. In order to achieve the optimal operation of distributed photovoltaic systems and improve energy utilization efficiency, accurate prediction of distributed photovoltaic power is essential. Distributed photovoltaic power prediction can help users reasonably plan electricity consumption, reduce grid load fluctuations, and optimize the operating efficiency of photovoltaic systems.

[0003] However, at present, my country generally uses administrative divisions to predict meteorological data, which deviates from my country's complex terrain and changeable weather conditions. Due to the influence of terrain, weather conditions in different locations within the same administrative region may vary greatly. At present, power generation forecasts based on regional meteorological data are inaccurate. Summary of the invention

[0004] Based on this, it is necessary to provide a distributed photovoltaic power prediction method, device, equipment and readable storage medium that can improve the accuracy of power prediction in response to the above technical problems.

[0005] In a first aspect, the present application provides a distributed photovoltaic power prediction method, comprising:

[0006] Obtain power generation data of each power station in the region;

[0007] Clustering multiple power stations based on power generation data to determine the power station category to which each power station belongs;

[0008] For each power station category, determine the target irradiance data collected by the target meteorological data collection station corresponding to the power station category;

[0009] The power generation power of the power station category is predicted based on the target irradiance data to obtain the target predicted power.

[0010] In one embodiment, the above-mentioned prediction of the power generation power of the power station category based on the target irradiance data to obtain the target power generation power includes:

[0011] Based on the target irradiance data, multiple power prediction models are used to predict the power generation power of the power station category to obtain the target predicted power.

[0012] In one embodiment, based on the target irradiance data, multiple power prediction models are used to predict the power generation power of the power station category to obtain the target predicted power, including:

[0013] Inputting the target irradiance data into a plurality of power prediction models respectively to obtain a plurality of initial predicted powers;

[0014] The target predicted power is determined based on the pre-acquired weight values ​​corresponding to the power prediction models and a plurality of initial predicted powers.

[0015] In one embodiment, the method further comprises:

[0016] Calculate the accuracy rates corresponding to multiple power prediction models respectively;

[0017] Based on the multiple accuracy rates, weight allocation processing is performed on the multiple power prediction models to obtain weight values ​​corresponding to the power prediction models.

[0018] In one embodiment, the weight allocation process is performed on multiple power prediction models based on multiple accuracy rates to obtain weight values ​​corresponding to each power prediction model, including:

[0019] Performing weight allocation processing on multiple power prediction models based on multiple accuracy rates to obtain initial weight values ​​corresponding to each power prediction model;

[0020] The initial weight value is corrected to obtain the target weight value.

[0021] In one embodiment, the weight allocation process is performed on multiple power prediction models based on multiple accuracy rates to obtain weight values ​​corresponding to each power prediction model, including:

[0022] Sort multiple accuracy rates and calculate the average accuracy rate corresponding to the multiple accuracy rates;

[0023] For a first power prediction model whose accuracy is less than the average accuracy, setting a target weight value of the first power prediction model to zero;

[0024] For the second power prediction models whose accuracy is greater than or equal to the average accuracy, the target weight values ​​of the second power prediction models are weighted in descending order of accuracy using a descending allocation method to obtain target weight values ​​corresponding to multiple second power prediction models.

[0025] In one embodiment, the target irradiance data collected by the target meteorological data collection site corresponding to the power station category is determined, including:

[0026] Calculate the average power value of all power plants in the power plant category;

[0027] For each candidate meteorological data collection site, calculate the average irradiance value of multiple candidate irradiance data collected by each candidate meteorological data collection site;

[0028] Based on the average power value and the average irradiance value of each candidate meteorological data collection site, a target meteorological data collection site is determined from multiple candidate meteorological data collection sites, and the candidate irradiance data of the target meteorological data collection site is determined as the target irradiance data.

[0029] In a second aspect, the present application also provides a distributed photovoltaic power prediction device, comprising:

[0030] A data acquisition module is used to obtain the power generation data of each power station in the area;

[0031] A power station category determination module, used to perform clustering processing on multiple power stations based on power generation data to determine the power station category to which each power station belongs;

[0032] An irradiance data determination module is used to determine, for each power station category, target irradiance data collected by a target meteorological data collection station corresponding to the power station category;

[0033] The power prediction module is used to predict the power generation power of the power station category based on the target irradiance data to obtain the target predicted power.

[0034] In a third aspect, the present application further provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the following steps are implemented:

[0035] Obtain power generation data of each power station in the region;

[0036] Clustering multiple power stations based on power generation data to determine the power station category to which each power station belongs;

[0037] For each power station category, determine the target irradiance data collected by the target meteorological data collection station corresponding to the power station category;

[0038] The power generation power of the power station category is predicted based on the target irradiance data to obtain the target predicted power.

[0039] In a fourth aspect, the present application further provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the following steps are implemented:

[0040] Obtain power generation data of each power station in the region;

[0041] Clustering multiple power stations based on power generation data to determine the power station category to which each power station belongs;

[0042] For each power station category, determine the target irradiance data collected by the target meteorological data collection station corresponding to the power station category;

[0043] The power generation power of the power station category is predicted based on the target irradiance data to obtain the target predicted power.

[0044] In a fifth aspect, the present application further provides a computer program product, including a computer program, which implements the following steps when executed by a processor:

[0045] Obtain power generation data of each power station in the region;

[0046] Clustering multiple power stations based on power generation data to determine the power station category to which each power station belongs;

[0047] For each power station category, determine the target irradiance data collected by the target meteorological data collection station corresponding to the power station category;

[0048] The power generation power of the power station category is predicted based on the target irradiance data to obtain the target predicted power.

[0049] The above-mentioned distributed photovoltaic power prediction method, device, equipment and readable storage medium first obtain the power generation data of each power station in the area; then, cluster multiple power stations based on the power generation data to determine the power station category to which each power station belongs; for each power station category, determine the target irradiance data collected by the target meteorological data collection station corresponding to the power station category; finally, predict the power generation of the power station category based on the target irradiance data to obtain the target predicted power; compared with the traditional method, the present application does not predict the power generation based on the meteorological data divided by the region, but clusters the power stations based on the power generation data, and classifies the power stations with similar power generation characteristics into one category. Then, determine the target meteorological data collection station and its target irradiance data that are most relevant to the power station of this category, and predict the power generation based on these target irradiance data, so that the selected meteorological data and the power generation of the power station have a high correlation, thereby significantly improving the accuracy of power prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] In order to more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the drawings required for use in the embodiments of the present application or related technical descriptions will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0051] Figure 1 An application environment diagram of a distributed photovoltaic power prediction method in an embodiment;

[0052] Figure 2 A schematic diagram of a process of a distributed photovoltaic power prediction method in one embodiment;

[0053] Figure 3 is a schematic flow chart of a distributed photovoltaic power prediction method in another embodiment;

[0054] Figure 4 is a schematic flow chart of a distributed photovoltaic power prediction method in another embodiment;

[0055] Figure 5 is a schematic flow chart of a distributed photovoltaic power prediction method in another embodiment;

[0056] Figure 6 is a schematic flow chart of a distributed photovoltaic power prediction method in another embodiment;

[0057] Figure 7 is a schematic flow chart of a distributed photovoltaic power prediction method in another embodiment;

[0058] Figure 8 is a structural block diagram of a distributed photovoltaic power prediction device in one embodiment;

[0059] Fig. 9 FIG. 4 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0060] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0061] The distributed photovoltaic power prediction method provided in the embodiment of the present application can be applied to Figure 1 In the application environment shown. Among them, the management system 101 communicates with the distributed photovoltaic power station 102 (hereinafter referred to as the power station 102) and the meteorological data collection site 103 through the network. The power station 102 needs to install monitoring equipment, such as power meters, energy meters and data acquisition systems. These monitoring equipment can record the power generation data of the power station 102 in real time and transmit these data to the management system 101 through the communication network. The data acquisition system can be a local server or a cloud platform, which is used to centrally store and manage data from different power stations 102.

[0062] First, the power generation data of each power station 102 is collected through the monitoring equipment of the power station 102, and then the power generation data is transmitted to the management system 101 using a communication network (such as the Internet, a dedicated network or wireless communication). In the management system 101, the power generation data is stored and managed. At the same time, the management system 101 also obtains irradiance data from each meteorological data collection station 103. Based on the irradiance data, the management system 101 predicts the power generation of the category to which the power station 102 belongs.

[0063] The power generation data refers to the value of the actual power output generated by the photovoltaic power station 102 in a specific time period. These power generation data are usually expressed in the form of power (such as kilowatts, kW) and will change over time. The power generation data includes the following key elements:

[0064] (1) Timestamp: The specific time point at which the power data is recorded, usually in hours, minutes or even seconds. For example, the average power generated per hour, or more fine-grained data per minute.

[0065] (2) Power value: the actual power output generated by the power station 102 at a specific time point or time period. For example, the power generated in a certain time period is 100 kW.

[0066] (3) Power station identification: Each data point is associated with a specific photovoltaic power station 102 to distinguish the power generation conditions of different power stations.

[0067] The meteorological data collection site 103 can provide meteorological data related to the category of the power station 102. The meteorological data collection site 103 usually monitors meteorological parameters such as irradiance, temperature, humidity, etc.

[0068] Irradiance data refers to the recorded solar radiation intensity. Irradiance data is usually expressed in irradiance per unit time (such as watts per square meter, W / m²).

[0069] In an exemplary embodiment, Figure 2 As shown in the figure, a distributed photovoltaic power prediction method is provided, which is applied to Figure 1 The management system 101 in FIG. 1 is taken as an example to illustrate the method, which includes the following steps 201 to 204. Among them:

[0070] Step 201, obtaining power generation data of each power station in the region.

[0071] Among them, "region" refers to a collection of multiple distributed photovoltaic power stations within a relatively concentrated geographical area. This area can be a town, a village, or even a specific area of ​​a smaller range. These photovoltaic power stations generate electricity through solar photovoltaic panels, and their power generation will be affected by local meteorological conditions (such as irradiance, temperature, wind speed, etc.).

[0072] In an embodiment of the present application, each power station in the region transmits the collected power generation data to the management system via the Internet, a dedicated network or wireless communication.

[0073] After acquiring the power generation data of each power station, the management system systematically stores and manages the power generation data.

[0074] It should be noted that the management system also provides a data access interface or platform, such as an application programming interface (API), to facilitate users to obtain and analyze the stored power generation data.

[0075] Step 202: cluster multiple power stations based on the power generation data to determine the power station category to which each power station belongs.

[0076] In an embodiment of the present application, the management system uses a clustering algorithm to cluster multiple power stations based on power generation data to determine the power station category to which each power station belongs.

[0077] In some embodiments, an unsupervised clustering algorithm (such as K-means clustering) can be used to cluster multiple power stations. First, select an appropriate number of clusters k as the number of initial cluster center points. Then randomly select k sample points as the initial cluster center points, and then for each remaining sample, divide it into the cluster corresponding to the most similar cluster center according to its distance from each cluster center. Then, recalculate the average value of all samples in each cluster center as the new cluster center. This process will be repeated until the position of the cluster center no longer changes, that is, the loss function (the sum of squared errors of the sample distance to the cluster center point, SSE) converges.

[0078] The specific steps are as follows:

[0079] (1) Selecting the initial cluster center point: Randomly select the power generation data of k power stations as the initial cluster center point.

[0080] (2) Assign samples to clusters: Calculate the distance between each power station and all cluster centers (using Euclidean distance), and assign the power station to the cluster corresponding to the cluster center closest to it.

[0081] (3) Update cluster center: For each cluster, recalculate the average value of the power generation data of all power plants in the cluster as the new cluster center.

[0082] (4) Iterative calculation: Repeat the steps of assigning samples and updating cluster centers until the cluster centers no longer change, that is, the loss function SSE converges.

[0083] In some embodiments, a hierarchical clustering algorithm can also be used to determine the power plant category to which each power plant belongs. First, the distance between each power plant is calculated to generate a distance matrix. Then, the agglomeration method in the hierarchical clustering algorithm is applied, and each power plant is considered as a separate cluster during initialization. The distance between each pair of clusters is calculated, and the two clusters with the closest distance are merged using the minimum distance, maximum distance, or average distance as the linking method, and the distance matrix is ​​updated. This process is repeated until all power plants are merged into one cluster, or a pre-set number of clusters is reached. By generating a dendrogram, the hierarchical structure of the clustering can be visualized, and the appropriate level can be selected as needed to determine the final number of clusters, thereby classifying the power plants.

[0084] Step 203 : for each power station category, determine the target irradiance data collected by the target meteorological data collection site corresponding to the power station category.

[0085] In the embodiment of the present application, first, the category to which each power station belongs is identified based on the previous cluster analysis results. For each power station category, a target meteorological data collection site corresponding to the power station category is selected. Then, the target irradiance data of the selected target meteorological data collection site is obtained.

[0086] Step 204 , predicting the power generation of the power station category based on the target irradiance data to obtain the target predicted power.

[0087] In the embodiment of the present application, the power generation power of the power station category is predicted based on the target irradiance data to obtain the target predicted power. For example, a support vector machine regression method can be used.

[0088] First, the target irradiance data and power generation data of power station categories are collected from weather stations, monitoring equipment or historical records. Next, features are extracted from the target irradiance data, including daily average, weekly average, maximum value, minimum value and standard deviation.

[0089] Subsequently, matching is performed through date or timestamp to associate the extracted features with the power generation data of the power plant category, ensuring that each target irradiance data has corresponding power generation and power plant category information.

[0090] Next, the extracted features and the corresponding power generation data are integrated into a unified table to form the final training data set. In the training data set, each row represents an observation sample, and the columns include the date, power station category, various irradiance features and their corresponding power generation.

[0091] Next, select the support vector machine regression model as the prediction tool. After setting the appropriate kernel function and hyperparameters, use the training data to train the model. After the model training is completed, divide the data set into a training set and a test set. When evaluating the prediction performance of the model on the test set, indicators such as mean square error and mean absolute error can be used to judge the accuracy of the model.

[0092] Finally, the trained support vector machine regression model is used to predict the new target irradiance data to obtain the predicted power generation value of the corresponding power station category.

[0093] In the above-mentioned distributed photovoltaic power prediction method, the power generation data of each power station in the area is first obtained; then, based on the power generation data, multiple power stations are clustered to determine the power station category to which each power station belongs; for each power station category, the target irradiance data collected by the target meteorological data collection station corresponding to the power station category is determined; finally, the power generation of the power station category is predicted based on the target irradiance data to obtain the target predicted power; compared with the traditional method, the present application does not predict the power generation based on the meteorological data divided by the region, but clusters the power stations based on the power generation data, and classifies the power stations with similar power generation characteristics into one category. Then, the target meteorological data collection station and its target irradiance data that are most relevant to the power station of this category are determined, and the power generation is predicted based on these target irradiance data, so that the selected meteorological data and the power generation of the power station have a high correlation, thereby significantly improving the accuracy of power prediction.

[0094] In an exemplary embodiment, step 204 includes:

[0095] Based on the target irradiance data, multiple power prediction models are used to predict the power generation power of the power station category to obtain the target predicted power.

[0096] In the embodiment of the present application, first, an appropriate power prediction model is selected for power prediction. The power prediction model may include linear regression, random forest regression, support vector machine regression, K nearest neighbor regression, ARIMA time series analysis model, artificial neural network, etc. (The following embodiment is described by taking 6 power prediction models as an example) These power prediction models can map the target irradiance data with the power generation data of the power station category, thereby predicting the power generation.

[0097] After the power prediction model is selected, it needs to be trained. The historical power data of the power station category is used as training data, and the corresponding target irradiance data is used as input features. The training process will adjust the parameters to minimize the error between the predicted power and the actual power.

[0098] After the training is completed, the prediction performance of the power prediction model is evaluated through cross-validation or validation set. Finally, the trained power prediction model is used for target prediction, that is, the target irradiance data is input into the model to obtain the predicted power generation power of the power station category.

[0099] In some embodiments, it is also possible to consider using model fusion technology to combine multiple prediction results to further improve the prediction accuracy.

[0100] In the above embodiment, multiple power prediction models are used for prediction, which can improve robustness; when some power prediction models perform poorly under certain circumstances, other power prediction models can still provide relatively accurate prediction results. By integrating the prediction results of different power prediction models, the impact of individual model prediction errors on the overall results can be reduced, thereby improving the stability and reliability of power prediction.

[0101] In an exemplary embodiment, Figure 3 As shown, step 204 specifically includes:

[0102] Step 301 : inputting target irradiance data into a plurality of power prediction models respectively to obtain a plurality of initial predicted powers.

[0103] In the embodiment of the present application, after the target irradiance data is obtained, it is input into a plurality of pre-trained power prediction models respectively. Each model generates an initial predicted power according to the input candidate irradiance data.

[0104] For example, linear regression model: input target irradiance characteristic data, calculate and output initial predicted power P1. Random forest regression model: input target irradiance characteristic data, calculate and output initial predicted power P2. Support vector machine regression model: input target irradiance characteristic data, calculate and output initial predicted power P3. K nearest neighbor regression model: input target irradiance characteristic data, calculate and output initial predicted power P4. ARIMA time series model: input target irradiance characteristic data, calculate and output initial predicted power P5. Artificial neural network model: input target irradiance characteristic data, calculate and output initial predicted power P6. Through the above steps, multiple initial predicted powers can be obtained, namely P1, P2, P3, P4, P5 and P6.

[0105] Among them, the linear regression model is a basic regression analysis method, which mainly fits the linear relationship between the input features and the output variables. When applied to the target irradiance characteristic data, it is first necessary to collect the input features related to power, such as light intensity, temperature, and time. Then, the regression coefficient is determined by the least squares method to establish a linear equation. After the linear regression model training is completed, the initial predicted power P1 can be calculated by inputting the new target irradiance characteristic data.

[0106] The random forest regression model is an integrated learning method that improves the prediction ability of the model by building multiple decision trees. When processing the target irradiance feature data, the data set is first randomly divided into multiple subsets, and a decision tree is trained for each subset. Each tree independently predicts the power value, and the final prediction result is obtained by averaging the output of all decision trees to obtain the initial predicted power P2.

[0107] The support vector machine regression model is a supervised learning algorithm based on margin maximization. When applied to target irradiance feature data, the support vector machine regression model divides the data into different categories by constructing a hyperplane. For regression problems, the support vector machine regression model will try to ensure that most data points are within the set tolerance while maximizing the distance between support vectors. During the training process, the support vector machine regression model will output the initial predicted power P3 based on the nonlinear transformation of the input features.

[0108] The K nearest neighbor regression model is an instance-based learning method. Its basic idea is to make predictions by measuring the distances between different data points. When processing target irradiance feature data, the K nearest neighbor regression model first determines a hyperparameter K, which represents the number of nearest neighbors considered when making predictions. For new input data, the K nearest neighbor regression model will find the K closest points to it and average the power values ​​of these points to obtain the initial predicted power P4.

[0109] The ARIMA time series model is a tool for time series data analysis, suitable for capturing trends and seasonality in data. When applied to target irradiance characteristic data, it is first necessary to perform a stationarity test and differential processing on the data to eliminate trend components. Subsequently, the appropriate AR and MA orders are determined based on the ACF and PACF graphs to construct the ARIMA time series model. By modeling historical power data, the ARIMA time series model can predict future power output and derive the initial predicted power P5.

[0110] When artificial neural networks analyze target irradiance characteristic data, they usually use a multi-layer perceptron structure, including input layer, hidden layer and output layer. The input layer receives irradiance characteristics, the hidden layer processes data through nonlinear activation functions, and the output layer generates the initial predicted power P6. Through the back propagation algorithm, the network gradually adjusts the weights to minimize the prediction error.

[0111] Step 302: Determine the target predicted power based on the pre-acquired weight values ​​corresponding to the power prediction models and a plurality of initial predicted powers.

[0112] In the embodiment of the present application, first, after obtaining multiple initial predicted powers, in order to integrate these prediction results into a more accurate final prediction value, it is necessary to use the weight values ​​of the predetermined power prediction model, and set these weights in turn to these weight values ​​respectively. , , , , and .

[0113] Exemplarily, first, the mean square error of the linear regression model, random forest regression model, support vector machine regression model, K nearest neighbor regression model, ARIMA time series model and artificial neural network model are calculated respectively. For example, after evaluation, the mean square error of the artificial neural network model is the lowest, which is 4, indicating that its prediction effect is the best; followed by the random forest regression model, with a mean square error of 5; followed by the ARIMA time series model, with a mean square error of 6; the mean square error of the support vector machine model is 7; the mean square error of the linear regression model is 10, and finally the mean square error of the K nearest neighbor regression model is the highest, which is 12.

[0114] Based on these mean square errors, the models are ranked from best to worst: artificial neural network model, random forest regression model, ARIMA time series model, support vector machine regression model, linear regression model, and K nearest neighbor regression model. To reflect the importance of each model in the final prediction, they are assigned corresponding weights. The best performing artificial neural network model receives the highest weight of 6, the random forest regression model receives a weight of 5, the ARIMA time series model receives a weight of 4, and the poorly performing K nearest neighbor regression model receives the lowest weight of 1.

[0115] Next, the initial prediction power of each model is weighted averaged using a predetermined weight value to determine the target prediction power. The process of determining the target prediction power can be shown in the following formula (1):

[0116] (1)

[0117] in, Predict power for the target.

[0118] In the above embodiment, the prediction results of multiple power prediction models can significantly improve the overall prediction accuracy. Each power prediction model has different capture capabilities for different data patterns and features. Through weight allocation, the advantages of each power prediction model can be maximized, thereby improving the accuracy of the final prediction result.

[0119] In an exemplary embodiment, Figure 4 As shown, the method may also include:

[0120] Step 401, respectively calculating the accuracy rates corresponding to a plurality of power prediction models.

[0121] In the embodiment of the present application, the accuracy rates corresponding to the multiple power prediction models are calculated respectively.

[0122] In some embodiments, each power prediction model is predicted using a test data set to obtain a predicted power value of each power prediction model. Next, a prediction error of each power prediction model is calculated. Then, the accuracy of each power prediction model is calculated based on the prediction error, which is usually defined as 1 minus the normalized prediction error.

[0123] The test dataset is used to evaluate the performance of the power prediction model in practical applications. Unlike the training dataset and validation dataset, the test dataset does not participate in the training or parameter adjustment of the model. Its purpose is to test the generalization ability and prediction accuracy of the model.

[0124] The test dataset usually contains features and target variables similar to the actual situation. In the power prediction scenario, the test dataset may include the following:

[0125] (1) Target irradiance characteristic data: such as light intensity, temperature, humidity, wind speed, etc. at different time points.

[0126] (2) Historical power output: the actual measured power value.

[0127] By analyzing the test dataset, we can evaluate the performance of each model under different weather conditions and light intensities.

[0128] The prediction error refers to the difference between the predicted value of the power prediction model and the actual observed value. It reflects the accuracy and reliability of the power prediction model in data prediction. The smaller the prediction error, the stronger the prediction ability of the model. When calculating the prediction error of the power prediction model, the absolute value of the difference between each predicted value and the actual value is calculated, and this absolute value is the prediction error.

[0129] For example, for the root mean square error and installed capacity, the error value of the accuracy power prediction model can be calculated by formula (2). Formula (2):

[0130] (2)

[0131] Among them, RMSE is the error value, PMi is the actual power at time i, PPi is the predicted power at time i, Cap is the installed capacity of the power station, and n is the number of samples. The accuracy is r=(1-RMSE)*100%. The smaller r is, the higher the accuracy, that is, the closer the predicted value is to the actual value, the better the prediction effect is.

[0132] Step 402 : performing weight allocation processing on multiple power prediction models based on multiple accuracy rates to obtain weight values ​​corresponding to each power prediction model.

[0133] In an embodiment of the present application, weight allocation processing is performed on multiple power prediction models based on multiple accuracy rates to obtain weight values ​​corresponding to each power prediction model.

[0134] In some embodiments, after determining the accuracy rates corresponding to multiple power prediction models, the accuracy rates of the various power prediction models are normalized so that the sum of their accuracy rates is 1, so that the accuracy rate of each power prediction model can be directly used as its weight.

[0135] The normalization process is to divide the accuracy of each power prediction model by the sum of the accuracy of all power prediction models. The specific steps are as follows:

[0136] (1) Calculate the total accuracy: Add up the accuracy of all power prediction models to get the total.

[0137] (2) Calculate the normalized accuracy: The normalized accuracy of each power prediction model is equal to the accuracy of the power prediction model divided by the total.

[0138] For example, if the accuracy of power prediction models A, B, and C are 0.8, 0.7, and 0.6 respectively, the total is 2.1. The normalized accuracy is:

[0139] Power prediction model A: 0.8 / 2.1≈0.38

[0140] Power prediction model B: 0.7 / 2.1≈0.33

[0141] Power prediction model C: 0.6 / 2.1≈0.29

[0142] Through such processing, the normalized sum of the accuracy is 1, which can effectively reflect the relative contribution of each power prediction model in the final prediction.

[0143] Next, verify whether the assigned weights are reasonable and make fine adjustments based on actual conditions and business needs so that the weights can reflect the prediction performance of each power prediction model.

[0144] In the above embodiment, since different power prediction models perform differently, weights are assigned to multiple power models based on accuracy. Even if some prediction models perform poorly, the contribution of other power prediction models with better performance to the final power prediction results can be improved through weight allocation, thereby improving the stability of the power prediction results.

[0145] In an exemplary embodiment, Figure 5 As shown, step 402 of "performing weight allocation processing on multiple power prediction models based on multiple accuracy rates to obtain weight values ​​corresponding to each power prediction model" may include:

[0146] Step 501 : performing weight allocation processing on multiple power prediction models based on multiple accuracy rates to obtain initial weight values ​​corresponding to each power prediction model.

[0147] In the embodiment of the present application, the accuracy of all power prediction models is added up and the sum between them is calculated. After the sum is obtained, the accuracy of each power prediction model is divided by the sum to obtain the initial weight value. These initial weight values ​​will reflect the relative performance of each power prediction model. For example, if the accuracy of power prediction model A is 0.80, power prediction model B is 0.70, and power prediction model C is 0.60, then their initial weight values ​​are 0.381, 0.333 and 0.286 respectively.

[0148] Step 502, correcting the initial weight value to obtain the target weight value.

[0149] In the embodiment of the present application, first, the initial weight values ​​are evaluated and the power prediction models with poor performance are identified. Assume that the initial weight value of the power prediction model A is 0.4, the initial weight value of the power prediction model B is 0.35, and the initial weight value of the power prediction model C is 0.25.

[0150] Next, these initial weight values ​​are modified. For example, if the power prediction model C performs poorly, its weight can be reduced, for example, to 0.2. At the same time, the reduced weight is allocated to other power prediction models with better performance. Assume that the 0.05 weight of the power prediction model C is allocated to the power prediction model A and the power prediction model B, and each power prediction model is increased by 0.025.

[0151] Then, update the initial weight value to get the target weight value. The target weight value is:

[0152] The target weight value of power prediction model A = 0.4 + 0.025 = 0.425;

[0153] The target weight value of power prediction model B = 0.35 + 0.025 = 0.375;

[0154] The target weight value of the power prediction model C = 0.2.

[0155] In the above embodiment, the corrected target weight value can more accurately reflect the performance of each power prediction model in a specific scenario. By identifying the power prediction model with poor performance and reducing its weight, it can be ensured that the final prediction result mainly depends on those power prediction models that perform well under various conditions, thereby improving the accuracy of the overall prediction. In addition, by dynamically adjusting the weight, the fluctuation of the power prediction model in the face of different data distributions or feature conditions can be reduced, making the prediction results more consistent and reliable.

[0156] In an exemplary embodiment, Figure 6 As shown, step 402 of "performing weight allocation processing on multiple power prediction models based on multiple accuracy rates to obtain weight values ​​corresponding to each power prediction model" may include:

[0157] Step 601, sorting multiple accuracy rates and calculating the average accuracy rate corresponding to the multiple accuracy rates.

[0158] In an embodiment of the present application, after obtaining the accuracy corresponding to each power prediction model, the accuracy of each power prediction model is sorted from large to small, and the average value of the accuracy of all power prediction models, that is, the average accuracy, is calculated.

[0159] Step 602: For a first power prediction model whose accuracy is less than the average accuracy, a target weight value of the first power prediction model is set to zero.

[0160] In an embodiment of the present application, for a model whose accuracy is less than the average value, its target weight value is set to zero, which indicates that the prediction result of the first power prediction model will not be taken into account.

[0161] Step 603: For the second power prediction models whose accuracy is greater than or equal to the average accuracy, the target weight values ​​of the second power prediction models are weighted in descending order of accuracy using a descending allocation method to obtain target weight values ​​corresponding to multiple second power prediction models.

[0162] In an embodiment of the present application, for the second power prediction model whose accuracy is greater than or equal to the average value, the weight is allocated in descending order according to the accuracy. The target weight value is allocated according to a predefined decreasing rule. For example, the second power prediction model with the highest accuracy has the highest weight, and the weights of subsequent second power prediction models decrease in sequence. Finally, verify whether the allocated weight is reasonable, and make fine adjustments according to the actual situation and business needs so that the weight can reflect the prediction performance of each power prediction model.

[0163] It should be noted that it is necessary to ensure that the sum of the target weight values ​​of all power models is 1 in order to perform weighted averaging. If necessary, the weights are normalized.

[0164] It should be noted that there is no sequential relationship between step 602 and step 603. In the actual implementation process, after sorting multiple accuracies and calculating the average accuracies corresponding to the multiple accuracies, for the first power prediction model whose accuracy is less than the average accuracy, execute step 602; for the second power prediction model whose accuracy is greater than or equal to the average accuracy, execute step 603.

[0165] In the above embodiment, by sorting and weighting, the power prediction model with excellent performance can occupy a larger proportion in the final prediction result. This method can maximize the advantages of the high-accuracy model, thereby improving the accuracy of the overall prediction. The weighting allocation strategy makes the high-accuracy model contribute more to the final result, and the low-accuracy model contributes less.

[0166] In an exemplary embodiment, Figure 7 As shown, step 203 "determining the target irradiance data collected by the target meteorological data collection site corresponding to the power station category" may include:

[0167] Step 701, calculating the average power values ​​of all power stations in the power station category.

[0168] In the embodiment of the present application, first, the power generation data of all power stations in the power station category are collected, and these data are averaged to obtain the average power value of the power station category.

[0169] Step 702: For each candidate meteorological data collection site, calculate the average irradiance value of multiple candidate irradiance data collected by each candidate meteorological data collection site.

[0170] In an embodiment of the present application, candidate irradiance data of each candidate meteorological data collection site is collected, and these candidate irradiance data are averaged to obtain an average irradiance value of multiple candidate irradiance data collected by each candidate meteorological data collection site.

[0171] Step 703, based on the average power value and the average irradiance value of each candidate meteorological data collection site, determine a target meteorological data collection site from multiple candidate meteorological data collection sites, and determine the candidate irradiance data of the target meteorological data collection site as the target irradiance data.

[0172] In the embodiment of the present application, a correlation analysis is performed based on the average power value of the power station category and the average irradiance value of each candidate meteorological data collection site to determine the target meteorological data collection site with the highest correlation. Finally, the target meteorological data collection site with the highest correlation is selected as the target meteorological data collection site, and its candidate irradiance data is determined as the target irradiance data.

[0173] In some embodiments, a statistical method (such as a linear correlation coefficient) may be used to analyze the relationship between the power generation data of each power station category and the candidate irradiance data of the selected meteorological station. Through this analysis, the strength of the correlation between the candidate irradiance data and the power generation of the power station category may be determined.

[0174] Based on the correlation analysis results, the target meteorological data collection site with the strongest correlation with the power generation data of each power station category is determined.

[0175] For example, suppose there are two power station categories, category A and category B. For category A, target meteorological data collection site X in the category area is selected and its candidate irradiance data is recorded. Then, the power generation data of the power station of category A is compared with the candidate irradiance data of the target meteorological data collection site X, and it is found that the correlation between the two is very strong. Therefore, the candidate irradiance data provided by the target meteorological data collection site X is determined as the target irradiance data of category A. Similarly, for the power station of category B, the target meteorological data collection site Y is selected as the target irradiance data.

[0176] In the above embodiment, by calculating the average power value of the power station category, the overall power generation situation of the power station of this category can be better represented. Combined with the average irradiation value, the most relevant meteorological data collection site can be matched more accurately, thereby improving the accuracy of the power prediction model.

[0177] Based on all the above embodiments, a distributed photovoltaic power prediction method is also provided, which includes:

[0178] Step 1: Obtain the power generation data of each power station in the area.

[0179] Step 2: cluster multiple power stations based on the power generation data to determine the power station category to which each power station belongs.

[0180] Step 3: For each power plant category, calculate the average power value of all power plants in the power plant category.

[0181] Step 4: For each candidate meteorological data collection site, calculate the average irradiance value of multiple candidate irradiance data collected by each candidate meteorological data collection site.

[0182] Step 5: Based on the average power value and the average irradiance value of each candidate meteorological data collection site, determine the target meteorological data collection site from multiple candidate meteorological data collection sites, and determine the candidate irradiance data of the target meteorological data collection site as the target irradiance data.

[0183] Step 6: Calculate the accuracy rates corresponding to the multiple power prediction models respectively.

[0184] Step 7: sort the multiple accuracy rates and calculate the average accuracy rate corresponding to the multiple accuracy rates.

[0185] Step 8: For the first power prediction model whose accuracy is less than the average accuracy, the target weight value of the first power prediction model is set to zero; for the second power prediction model whose accuracy is greater than or equal to the average accuracy, the target weight value of the second power prediction model is weighted in descending order of accuracy using a descending distribution method to obtain target weight values ​​corresponding to multiple second power prediction models.

[0186] Step 9: Input the target irradiance data into multiple power prediction models respectively to obtain multiple initial predicted powers.

[0187] Step 10: Determine the target predicted power based on the acquired weight values ​​corresponding to each power prediction model and a plurality of initial predicted powers.

[0188] It should be understood that, although the steps in the flowcharts involved in the above embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps is not strictly limited in order, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above embodiments may include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.

[0189] Based on the same inventive concept, the embodiment of the present application also provides a distributed photovoltaic power prediction device for implementing the above-mentioned distributed photovoltaic power prediction method. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above-mentioned method, so the specific limitations in one or more distributed photovoltaic power prediction device embodiments provided below can refer to the above-mentioned limitations on the distributed photovoltaic power prediction method, and will not be repeated here.

[0190] In an exemplary embodiment, Figure 8 As shown, a distributed photovoltaic power prediction device is provided, including: a data acquisition module 801, a power station category determination module 802, an irradiance data determination module 803 and a power prediction module 804, wherein:

[0191] The data acquisition module 801 is used to acquire the power generation data of each power station in the region;

[0192] A power station category determination module 802 is used to perform clustering processing on multiple power stations based on the power generation data to determine the power station category to which each power station belongs;

[0193] The irradiance data determination module 803 is used to determine, for each power station category, the target irradiance data collected by the target meteorological data collection site corresponding to the power station category;

[0194] The power prediction module 804 is used to predict the power generation power of the power station category based on the target irradiance data to obtain the target predicted power.

[0195] In an exemplary embodiment, another distributed photovoltaic power prediction device is provided, wherein the power prediction module 804 is specifically used for:

[0196] Based on the target irradiance data, multiple power prediction models are used to predict the power generation power of the power station category to obtain the target predicted power; the target irradiance data are respectively input into multiple power prediction models to obtain multiple initial predicted powers; based on the weight values ​​corresponding to each power prediction model obtained in advance and the multiple initial predicted powers, the target predicted power is determined.

[0197] In an exemplary embodiment, another distributed photovoltaic power prediction device is provided, and the device further includes:

[0198] An accuracy calculation module, used to calculate the accuracy corresponding to multiple power prediction models respectively;

[0199] The weight value determination module is used to perform weight allocation processing on multiple power prediction models based on multiple accuracy rates to obtain weight values ​​corresponding to each power prediction model.

[0200] In an exemplary embodiment, another distributed photovoltaic power prediction device is also provided, wherein the weight value determination module is specifically used to:

[0201] Performing weight allocation processing on multiple power prediction models based on multiple accuracy rates to obtain initial weight values ​​corresponding to each power prediction model;

[0202] The initial weight value is corrected to obtain the target weight value.

[0203] In an exemplary embodiment, another distributed photovoltaic power prediction device is also provided, and the above-mentioned weight value determination module is specifically used to: sort multiple accuracies and calculate the average accuracies corresponding to the multiple accuracies; for a first power prediction model whose accuracy is less than the average accuracy, set the target weight value of the first power prediction model to zero; for a second power prediction model whose accuracy is greater than or equal to the average accuracy, weight the target weight value of the second power prediction model is allocated in descending order of accuracy using a decreasing allocation method to obtain target weight values ​​corresponding to multiple second power prediction models.

[0204] In an exemplary embodiment, another distributed photovoltaic power prediction device is also provided, and the power prediction module 804 is specifically used to: calculate the average power value of all power stations in the power station category; for each candidate meteorological data collection site, calculate the average irradiance value of multiple candidate irradiance data collected by each candidate meteorological data collection site; based on the average power value and the average irradiance value of each candidate meteorological data collection site, determine the target meteorological data collection site from multiple candidate meteorological data collection sites, and determine the candidate irradiance data of the target meteorological data collection site as the target irradiance data

[0205] Each module in the above-mentioned distributed photovoltaic power prediction device can be implemented in whole or in part by software, hardware and a combination thereof. Each of the above-mentioned modules can be embedded in or independent of a processor in a computer device in the form of hardware, or can be stored in a memory in a computer device in the form of software, so that the processor can call and execute the operations corresponding to each of the above modules.

[0206] In an exemplary embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as shown in FIG. Fig. 9As shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. The processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store data in the distributed photovoltaic power prediction process. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a distributed photovoltaic power prediction method is implemented.

[0207] Those skilled in the art will understand that Fig. 9 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0208] In an exemplary embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the computer program, the following steps are implemented:

[0209] Obtain power generation data of each power station in the region;

[0210] Clustering multiple power stations based on power generation data to determine the power station category to which each power station belongs;

[0211] For each power station category, determine the target irradiance data collected by the target meteorological data collection station corresponding to the power station category;

[0212] The power generation power of the power station category is predicted based on the target irradiance data to obtain the target predicted power.

[0213] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0214] Based on the target irradiance data, multiple power prediction models are used to predict the power generation power of the power station category to obtain the target predicted power.

[0215] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0216] Inputting the target irradiance data into a plurality of power prediction models respectively to obtain a plurality of initial predicted powers;

[0217] The target predicted power is determined based on the pre-acquired weight values ​​corresponding to the power prediction models and a plurality of initial predicted powers.

[0218] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0219] Calculate the accuracy rates corresponding to multiple power prediction models respectively;

[0220] Based on the multiple accuracy rates, weight allocation processing is performed on the multiple power prediction models to obtain weight values ​​corresponding to the power prediction models.

[0221] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0222] Performing weight allocation processing on multiple power prediction models based on multiple accuracy rates to obtain initial weight values ​​corresponding to each power prediction model;

[0223] The initial weight value is corrected to obtain the target weight value.

[0224] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0225] Sort multiple accuracy rates and calculate the average accuracy rate corresponding to the multiple accuracy rates;

[0226] For a first power prediction model whose accuracy is less than the average accuracy, setting a target weight value of the first power prediction model to zero;

[0227] For the second power prediction models whose accuracy is greater than or equal to the average accuracy, the target weight values ​​of the second power prediction models are weighted in descending order of accuracy using a descending allocation method to obtain target weight values ​​corresponding to multiple second power prediction models.

[0228] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0229] Calculate the average power value of all power plants in the power plant category;

[0230] For each candidate meteorological data collection site, calculate the average irradiance value of multiple candidate irradiance data collected by each candidate meteorological data collection site;

[0231] Based on the average power value and the average irradiance value of each candidate meteorological data collection site, a target meteorological data collection site is determined from multiple candidate meteorological data collection sites, and the candidate irradiance data of the target meteorological data collection site is determined as the target irradiance data.

[0232] In one embodiment, a computer readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, the following steps are implemented:

[0233] Obtain power generation data of each power station in the region;

[0234] Clustering multiple power stations based on power generation data to determine the power station category to which each power station belongs;

[0235] For each power station category, determine the target irradiance data collected by the target meteorological data collection station corresponding to the power station category;

[0236] The power generation power of the power station category is predicted based on the target irradiance data to obtain the target predicted power.

[0237] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0238] Based on the target irradiance data, multiple power prediction models are used to predict the power generation power of the power station category to obtain the target predicted power.

[0239] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0240] Inputting target irradiance data into multiple power prediction models respectively to obtain multiple initial predicted powers;

[0241] The target predicted power is determined based on the pre-acquired weight values ​​corresponding to the power prediction models and a plurality of initial predicted powers.

[0242] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0243] Calculate the accuracy rates corresponding to multiple power prediction models respectively;

[0244] Based on the multiple accuracy rates, weight allocation processing is performed on the multiple power prediction models to obtain weight values ​​corresponding to the power prediction models.

[0245] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0246] Performing weight allocation processing on multiple power prediction models based on multiple accuracy rates to obtain initial weight values ​​corresponding to each power prediction model;

[0247] The initial weight value is corrected to obtain the target weight value.

[0248] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0249] Sort multiple accuracy rates and calculate the average accuracy rate corresponding to the multiple accuracy rates;

[0250] For a first power prediction model whose accuracy is less than the average accuracy, setting a target weight value of the first power prediction model to zero;

[0251] For the second power prediction models whose accuracy is greater than or equal to the average accuracy, the target weight values ​​of the second power prediction models are weighted in descending order of accuracy using a descending allocation method to obtain target weight values ​​corresponding to multiple second power prediction models.

[0252] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0253] Calculate the average power value of all power plants in the power plant category;

[0254] For each candidate meteorological data collection site, calculate the average irradiance value of multiple candidate irradiance data collected by each candidate meteorological data collection site;

[0255] Based on the average power value and the average irradiance value of each candidate meteorological data collection site, a target meteorological data collection site is determined from multiple candidate meteorological data collection sites, and the candidate irradiance data of the target meteorological data collection site is determined as the target irradiance data.

[0256] In one embodiment, a computer program product is provided, comprising a computer program, which, when executed by a processor, implements the following steps:

[0257] Obtain power generation data of each power station in the region;

[0258] Clustering multiple power stations based on power generation data to determine the power station category to which each power station belongs;

[0259] For each power station category, determine the target irradiance data collected by the target meteorological data collection station corresponding to the power station category;

[0260] The power generation power of the power station category is predicted based on the target irradiance data to obtain the target predicted power.

[0261] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0262] Based on the target irradiance data, multiple power prediction models are used to predict the power generation power of the power station category to obtain the target predicted power.

[0263] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0264] Inputting target irradiance data into multiple power prediction models respectively to obtain multiple initial predicted powers;

[0265] The target predicted power is determined based on the pre-acquired weight values ​​corresponding to the power prediction models and a plurality of initial predicted powers.

[0266] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0267] Calculate the accuracy rates corresponding to multiple power prediction models respectively;

[0268] Based on the multiple accuracy rates, weight allocation processing is performed on the multiple power prediction models to obtain weight values ​​corresponding to the power prediction models.

[0269] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0270] Performing weight allocation processing on multiple power prediction models based on multiple accuracy rates to obtain initial weight values ​​corresponding to each power prediction model;

[0271] The initial weight value is corrected to obtain the target weight value.

[0272] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0273] Sort multiple accuracy rates and calculate the average accuracy rate corresponding to the multiple accuracy rates;

[0274] For a first power prediction model whose accuracy is less than the average accuracy, setting a target weight value of the first power prediction model to zero;

[0275] For the second power prediction models whose accuracy is greater than or equal to the average accuracy, the target weight values ​​of the second power prediction models are weighted in descending order of accuracy using a descending allocation method to obtain target weight values ​​corresponding to multiple second power prediction models.

[0276] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0277] Calculate the average power value of all power plants in the power plant category;

[0278] For each candidate meteorological data collection site, calculate the average irradiance value of multiple candidate irradiance data collected by each candidate meteorological data collection site;

[0279] Based on the average power value and the average irradiance value of each candidate meteorological data collection site, a target meteorological data collection site is determined from multiple candidate meteorological data collection sites, and the candidate irradiance data of the target meteorological data collection site is determined as the target irradiance data.

[0280] A person of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment method can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, an artificial intelligence (AI) processor, etc., but are not limited to this.

[0281] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0282] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. A distributed photovoltaic power prediction method, characterized in that: The method comprises: Obtain power generation data of each power station in the region; Performing clustering processing on the plurality of power stations based on the power generation data to determine the power station category to which each of the power stations belongs; For each of the power station categories, determining the target irradiance data collected by the target meteorological data collection site corresponding to the power station category; The power generation power of the power station category is predicted based on the target irradiance data to obtain the target predicted power.

2. The method according to claim 1, characterized in that The step of predicting the power generation of the power station category based on the target irradiance data to obtain the target power generation includes: Based on the target irradiance data, multiple power prediction models are used to predict the power generation power of the power station category to obtain the target predicted power.

3. The method according to claim 2, characterized in that The method of using a plurality of power prediction models to predict the power generation power of the power station category based on the target irradiance data to obtain the target predicted power includes: Inputting the target irradiance data into the plurality of power prediction models respectively to obtain a plurality of initial predicted powers; The target predicted power is determined based on the pre-acquired weight values ​​corresponding to the power prediction models and the multiple initial predicted powers.

4. The method according to claim 3, characterized in that: The method further comprises: Calculating the accuracy rates corresponding to the multiple power prediction models respectively; Based on the multiple accuracy rates, weight allocation processing is performed on the multiple power prediction models to obtain a weight value corresponding to each power prediction model.

5. The method according to claim 4, characterized in that The step of performing weight allocation processing on the multiple power prediction models based on the multiple accuracy rates to obtain a weight value corresponding to each of the power prediction models includes: Performing weight allocation processing on the multiple power prediction models based on the multiple accuracy rates to obtain initial weight values ​​corresponding to the power prediction models; The initial weight value is corrected to obtain a target weight value.

6. The method according to claim 5, characterized in that The step of performing weight allocation processing on the multiple power prediction models based on the multiple accuracy rates to obtain a weight value corresponding to each of the power prediction models includes: Sorting the multiple accuracy rates, and calculating the average accuracy rate corresponding to the multiple accuracy rates; For a first power prediction model whose accuracy is less than the average accuracy, setting a target weight value of the first power prediction model to zero; For the second power prediction model whose accuracy is greater than or equal to the average accuracy, the target weight values ​​of the second power prediction model are weighted in descending order of accuracy using a descending allocation method to obtain target weight values ​​corresponding to multiple second power prediction models.

7. The method according to claim 1, characterized in that The determining of the target irradiance data collected by the target meteorological data collection site corresponding to the power station category includes: Calculate the average power value of all power stations in the power station category; For each candidate meteorological data collection site, calculating an average irradiance value of a plurality of candidate irradiance data collected by each candidate meteorological data collection site; Based on the average power value and the average irradiance value of each candidate meteorological data collection site, a target meteorological data collection site is determined from multiple candidate meteorological data collection sites, and the candidate irradiance data of the target meteorological data collection site is determined as the target irradiance data.

8. A distributed photovoltaic power prediction device, characterized in that: The device comprises: A data acquisition module is used to obtain the power generation data of each power station in the area; A power station category determination module, configured to perform clustering processing on the plurality of power stations based on the power generation data, and determine the power station category to which each of the power stations belongs; An irradiance data determination module, used for determining, for each power station category, target irradiance data collected by a target meteorological data collection site corresponding to the power station category; A power prediction module is used to predict the power generation power of the power station category based on the target irradiance data to obtain a target predicted power.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

11. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images