Distributed photovoltaic power generation power intelligent prediction system and method

Abstract

The invention discloses a distributed photovoltaic power generation power intelligent prediction system and method, and relates to the technical field of distributed photovoltaic power generation power prediction systems.The system comprises a weather classification module, a data processing module and an IPSO-LSTM module, and the weather classification module carries out generalized division on weather types based on historical environmental factor data; the data processing module carries out data normalization processing according to photovoltaic historical power generation data, and noise interference is reduced; and the IPSO-LSTM module uses the trained weather type IPSO-LSTM model to carry out distributed photovoltaic power generation power intelligent prediction by using the normalized power generation data according to the weather type to which the prediction day belongs. According to the system and the method, the influence of weather types on power generation power prediction is fully considered, the non-stationary influence in original power generation data is reduced, and compared with a conventional short-term photovoltaic power generation power prediction model, the system and the method have higher prediction precision.

Description

Technical Field

[0001] The present invention relates to the technical field of distributed photovoltaic power generation prediction systems, and in particular to a distributed photovoltaic power generation intelligent prediction system and method. Background Art

[0002] Distributed photovoltaic power generation is highly random and environmentally dependent, resulting in large fluctuations in its output power. After the photovoltaic power generation system is connected to the power grid, its volatility will significantly affect the stability of the power system. In order to ensure the safe operation of the power system, it is often necessary to take measures to abandon the light, resulting in a large amount of energy waste. Therefore, accurate photovoltaic power generation prediction is crucial to improving the stability of the power system and reducing energy waste.

[0003] Photovoltaic power generation is greatly affected by weather and has strong random fluctuations. The existing photovoltaic power generation prediction system focuses on modeling and training of neural network models such as LSTM, and has insufficient application in processing weather type changes and photovoltaic power generation historical data, resulting in large errors and low accuracy in photovoltaic power generation prediction.

[0004] Patent document with application number 202410281575.9 discloses a method and system for predicting the power generation of photovoltaic equipment. The scheme uses a reconstructed solar radiation intensity model to predict the radiation intensity and obtain the final radiation intensity result; the photovoltaic power generation is predicted based on the final radiation intensity result. The impact of salt fog on photovoltaic power generation is taken into account to improve the accuracy of photovoltaic power generation prediction. However, the scheme also has the problem of insufficient application of weather type changes and photovoltaic power generation historical data. Summary of the invention

[0005] In view of the above-mentioned problems, the purpose of the present invention is to provide a distributed photovoltaic power generation intelligent prediction system and method to solve the problem of insufficient application of weather type changes and photovoltaic power generation historical data, and improve the accuracy of photovoltaic power generation prediction.

[0006] The embodiments of the present invention provide a distributed photovoltaic power generation intelligent prediction system and method.

[0007] A first aspect: A distributed photovoltaic power generation intelligent prediction system, comprising: The weather classification module broadly divides weather types based on historical environmental factor data; The data processing module performs data normalization based on the historical photovoltaic power generation data to reduce noise interference; The IPSO-LSTM module uses the trained weather type IPSO-LSTM model to perform intelligent prediction of distributed photovoltaic power generation based on the weather type of the forecast day and the normalized power generation data.

[0008] A second aspect: A prediction method based on the above system, comprising the following steps: S1. Collect historical data on environmental factors of photovoltaic power generation, select influencing factors, and make a broad classification of weather types based on the influencing factors; S2. Collect historical data of photovoltaic power generation and normalize the historical power generation data; S3. Based on the broad classification of weather types, IPSO-LSTM models of different weather types are constructed and trained; S4. According to the weather type of the forecast day, select the IPSO-LSTM model of the corresponding weather type; S5. Selecting the corresponding normalized historical power generation data according to the weather type of the forecast day; S6. Based on the selected weather type IPSO-LSTM model, the selected normalized power generation data is used to predict the daily power generation power.

[0009] Furthermore, when selecting influencing factors in S1, the Pearson correlation coefficient is used, and the formula is: (1) in, y is the photovoltaic power generation power; α are the relevant meteorological variable factors; t is the Pearson correlation coefficient.

[0010] Furthermore, when the weather types are broadly divided according to the influencing factors in S1, the generalized weather classification is performed based on K-means++ clustering, and the steps include: S11. Select the required number of cluster centers. The formula is: (2) (3) in, X For the data set, t is the number of cluster centers currently selected, μ i For the i Cluster centers, D(x) is the distance between each point in the data set and the nearest point that has been selected as the cluster center; P(x) The distance of all points x is calculated D(x) After that, each point x The probability of being selected as the next cluster center.

[0011] S12. Assign each point to the nearest cluster C j ; S13, according to each cluster Cj The cluster center is updated with the mean value until the cluster center no longer changes or the change is less than the preset threshold; (4) Among them, | C j | for cluster C j The number of samples in .

[0012] Furthermore, when normalizing the historical power generation data in S2, the ICEEMDAN algorithm is used to perform multi-layer modal decomposition of the historical power generation data to obtain different modal components and residual components in each layer, thereby capturing the details and features of the historical power generation data and reducing noise interference.

[0013] Furthermore, the state update formula of the LSTM model in the IPSO-LSTM model is: (5) in, is the sigmoid function, tanh() is the hyperbolic tangent function, i t is the input gate, f t For the Gate of Forgetfulness, o t is the output gate, x t ，h t ，c t Respectively represent the input, output and cell state of LSTM at the current time step t, x t-1 ，h t-1 ，c t-1 Respectively represent the input, output and cell state of LSTM at time step t-1, W xi 、W hi 、W ci 、b i 、W xf 、W hf 、W cf 、b f 、W xc 、W hc 、b c 、W xo 、W ho 、W co and b o are model parameters.

[0014] Furthermore, the network parameters of the LSTM model are optimized using the IPSO algorithm to construct an IPSO-LSTM model.

[0015] A third aspect: An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the method provided in the second aspect when executing the program.

[0016] A fourth aspect: A non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method provided in the second aspect.

[0017] Beneficial effects of the present invention: In order to solve the problems that photovoltaic power generation is greatly affected by weather types and historical power generation data has strong random volatility, the present invention adopts the K-means++ clustering algorithm to divide the generalized weather types based on historical irradiance data, and then decomposes the photovoltaic power generation data into several modal components and residual components with different frequencies through the ICEEMDAN algorithm to reduce the non-stationarity of the original power generation data. Different IPSO-LSTM prediction models are constructed and trained under different weather types. Compared with conventional short-term photovoltaic power generation prediction models, the present invention has higher prediction accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 It is a structural schematic diagram of a distributed photovoltaic power generation intelligent prediction system of the present invention; Figure 2 A schematic diagram of a process of a distributed photovoltaic power generation intelligent prediction method of the present invention; Figure 3 This is a schematic diagram of the LSTM model optimized by the IPSO algorithm in the present invention; Figure 4 It is a schematic diagram of the structure of the electronic device of the present invention. DETAILED DESCRIPTION

[0019] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar symbols throughout represent the same or similar elements or elements with the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and cannot be understood as limiting the present invention.

[0020] Photovoltaic power generation is greatly affected by weather, and historical power generation data is subject to strong random fluctuations. The existing photovoltaic power generation prediction system does not adequately utilize weather type changes and historical photovoltaic power generation data, resulting in large prediction and measurement errors.

[0021] In view of the above problems, the present invention provides a distributed photovoltaic power generation intelligent prediction system. Figure 1 The present invention provides a schematic diagram of the structure of a distributed photovoltaic power generation intelligent prediction system, which includes a weather classification module, a data processing module, an IPSO-LSTM module, etc.

[0022] Among them, the weather classification module makes a broad classification of weather types based on historical environmental factor data; the data processing module reduces noise interference and performs data normalization based on historical photovoltaic power generation data; the IPSO-LSTM module uses the trained weather type IPSO-LSTM model to perform intelligent prediction of distributed photovoltaic power generation based on the weather type of the predicted day and the normalized power generation data.

[0023] like Figure 2 As shown, the present invention also discloses a distributed photovoltaic power generation intelligent prediction method, comprising: S1. Collect historical data on environmental factors of photovoltaic power generation, select influencing factors, and make a broad classification of weather types based on the influencing factors; S2. Collect historical data of photovoltaic power generation and normalize the historical power generation data; S3. Based on the broad classification of weather types, IPSO-LSTM models of different weather types are constructed and trained; S4. According to the weather type of the forecast day, select the IPSO-LSTM model of the corresponding weather type; S5. Selecting the corresponding normalized historical power generation data according to the weather type of the forecast day; S6. Based on the selected weather type IPSO-LSTM model, the selected normalized power generation data is used to predict the daily power generation power.

[0024] Different weather factors have different degrees of influence on the power generation of photovoltaic arrays. Therefore, when broadly classifying weather types, we first determine the factors that have the greatest impact on power generation.

[0025] To determine the most important factors affecting photovoltaic power generation, the Pearson correlation coefficient method can be used to analyze the correlation between each factor and power generation. The formula is:

[0026] in, y is the photovoltaic power generation power; α are the relevant meteorological variables. t is the Pearson correlation coefficient.

[0027] When 0≤t≤0.2 The correlation is very weak or irrelevant; 0.2≤t≤0.4 is weakly correlated; 0.4≤t≤0.6 The correlation is of moderate strength; 0.6≤t≤0.8 is strongly correlated; 0.8≤t≤1.0 Strongly correlated.

[0028] By calculating the correlation coefficient, the Pearson correlation coefficient between each influencing factor and photovoltaic power generation was obtained, as listed in Table 1.

[0029] Table 1 Pearson correlation coefficient of factors affecting photovoltaic power generation

[0030] It can be seen from Table 1 that there is a Pearson correlation between the irradiance, temperature, wind speed, humidity and wind direction in the environmental factors and the photovoltaic power generation power. It can be seen from Table 1 that there is a strong positive correlation between the irradiance and the photovoltaic power generation power, indicating that it has the greatest impact on the output power of the photovoltaic array. The photovoltaic power generation power also shows a certain positive correlation with the temperature and wind speed, and a negative correlation with humidity, but the impact is smaller than that of the irradiance. In addition, the correlation between the wind direction factor and the photovoltaic power generation power is extremely small. Therefore, generalized weather classification can be performed based on irradiance.

[0031] Then, we perform generalized weather type classification based on K-means++ clustering, including: Initialize the cluster center first, for example, randomly select a point from the data set as the first cluster center μ 1 , and then select the remaining cluster centers.

[0032] For each point in the data set, calculate its distance to the nearest point that has been selected as the cluster center. The formula is: (2) in, t is the number of cluster centers currently selected, μ i For the i Cluster centers, D(x)is the distance between each point in the data set and the nearest point that has been selected as the cluster center.

[0033] After calculating all the points x Distance D(x) Then calculate each point x The probability of being selected as the next cluster center is: (3) in, X For the data set, P(x) The distance of all points x is calculated D(x) After that, each point x The probability of being selected as the next cluster center.

[0034] According to the probability distribution of formula (3), a point is randomly selected from the data set as the next cluster center. Since the probability of selecting a farther point as the cluster center is higher, the distribution of cluster centers can be made more dispersed in the end. The above process is repeated until the required number of cluster centers are selected.

[0035] Then K-means++ clustering is performed, including: Assign each point to the nearest cluster, for each point in the dataset x i , calculate its difference with each cluster center μ j Euclidean distance , point x i Assign to the cluster corresponding to the nearest cluster center C j .

[0036] According to each cluster C j The mean of updates the cluster center, the formula is: (4) Among them, | C j | for cluster C j The number of samples in .

[0037] Repeat the above assignment and update steps until the cluster center no longer changes or the change is less than a preset threshold.

[0038] By using the K-means++ clustering method to divide different clusters, the historical environmental data of photovoltaic power generation can be classified into generalized weather types.

[0039] When making short-term power generation forecasts for photovoltaic arrays, it is also crucial to perform modal decomposition of historical power generation data. Through decomposition technology, the non-stationarity of historical power generation data signals can be effectively processed, the fluctuation components of different time scales can be extracted, and noise interference can be reduced. The processed historical power generation data can be used to improve the accuracy and stability of the prediction model.

[0040] When normalizing historical power generation data, the ICEEMDAN algorithm can be used. By adding adaptive Gaussian white noise, the non-stationary problem of historical power generation data can be more effectively handled. The ICEEMDAN algorithm uses adaptive Gaussian white noise to decompose the original photovoltaic power generation signal. X(t) It is decomposed into several IMF and Res functions with frequencies from high to low, thereby effectively alleviating the influence of residual noise and modal aliasing and reducing pseudo components in the decomposition process of power generation data.

[0041] The ICEEMDAN algorithm is used to analyze the original photovoltaic power signal X(t) The decomposition process is: (5) in, X i (t) For the construction i A noise signal; E 1 (∙) It is the operator that can generate the first-order modal component in the empirical mode decomposition algorithm; w i (t) For the i The Gaussian white noise with a mean of 0 and a variance of 1 is added. β 0 is the standard deviation of the noise when the signal is first decomposed.

[0042] The ICEEMDAN algorithm is used to decompose the noisy signal. X i (t) After calculating the local mean, we find the average value and get the first-order residual component, which is expressed as: (7) In the formula, M(·) Represents the local mean of the signal.

[0043] Calculate the first-order modal component from the original signal X(t) Subtracting the first-order residual, the expression is: (8) Calculate the second-order residual component and modal component, and take the former to estimate The average of the local means, and then subtract the second-order residual from the first-order residual, the expression is: (9) (10) Derived k The order residual components and modal components are expressed as: (11) (12) Repeat the above steps until all residual components and modal components are obtained.

[0044] The ICEEMDAN algorithm can be used to perform multi-layer modal decomposition on historical photovoltaic power generation data to obtain different modal components and residual components in each layer, capture data details and features, reduce noise interference, and reduce the non-stationarity of the original power generation data. The modal components and residual components of each layer are combined with the IPSO-LSTM model under different weather types for prediction, and all predicted value components are accumulated to improve the accuracy of power prediction.

[0045] According to the classification results of generalized weather types, different IPSO-LSTM prediction models are established under different weather types, and trained and predicted respectively.

[0046] Among them, the long short-term memory network (LSTM) is a special recurrent neural network (RNN), which is mainly used to process and predict time series data. Compared with RNN, LSTM has added memory units, which can better capture the dependencies with larger time step distances in the time series.

[0047] Simple RNNs will encounter the problem of gradient vanishing or exploding when modeling long sequence data. The LSTM model adds a memory unit c that can record historical information. This unit controls the inflow and outflow of information through three gates: input gate, forget gate, and output gate. Each of these three gates is a sigmoid layer, and the degree of its opening or closing determines whether the information can enter the memory unit; LSTM also uses a tanh layer to update the content of the memory unit; in this way, LSTM can transmit information between time steps and model long-distance dependencies more effectively.

[0048] LSTM has a unique advantage in model time correlation. The LSTM network is mainly composed of a memory unit and three gate units, namely input gate i, forget gate f and output gate o. When the time step is t, the mathematical formula for updating the LSTM model is: (5) in, is the sigmoid function, tanh() is the hyperbolic tangent function, i t is the input gate, f t For the Gate of Forgetfulness, o t is the output gate, x t ，h t ，c t Respectively represent the input, output and cell state of LSTM at the current time step t, x t-1 ，h t-1 ，c t-1 Respectively represent the input, output and cell state of LSTM at time step t-1, W xi 、W hi 、W ci 、b i 、W xf 、W hf 、W cf 、b f 、W xc 、W hc 、 b c 、W xo 、W ho 、W co and b o is the model parameter; the three gate control units are calculated using the sigmoid function σ to achieve 0-1 range control; the final output h t Combined with memory unit c t and output gate o t ; These three gating units can precisely control the flow of information and capture long-range correlations in time series.

[0049] The sigmoid gated activation function enables the LSTM network to learn complex nonlinear time series dynamic behaviors; applying LSTM to distributed photovoltaic power generation power forecasting can capture the correlation between the output at each moment and the past moments, and achieve high-precision short-term and ultra-short-term power forecasting of distributed photovoltaic systems.

[0050] Furthermore, if Figure 3 As shown in the figure, in order to accurately predict the power generation of the photovoltaic power generation system, the IPSO algorithm is used to optimize the LSTM network parameters. Specifically: The values ​​of certain hyperparameters in the LSTM neural network control the prediction performance of the network. In order to improve the prediction performance of the network, the IPSO algorithm is combined with the LSTM neural network model to establish the IPSO-LSTM prediction model.

[0051] like Figure 3 As shown in the figure, the model first uses the number of neurons in the first layer of the LSTM network s1, the number of neurons in the second layer s2, the learning rate e and the number of iterations m as the optimization targets of the IPSO algorithm, and randomly initializes the particle position information according to the value range of each parameter: then the LSTM neural network model is established using the parameters corresponding to the particle position, and the LSTM model is trained and predicted using the system's monitoring data; the mean square error of the model is used as the fitness value of the improved particle swarm algorithm, and the fitness function is defined as: (13) Where n is the number of test data; y i is the true value of the test data; is the predicted value of the test data.

[0052] By comparing the fitness value of each particle, the current individual extreme value can be obtained bpest i and the current global optimal solution gbest i . They are used to update the particle velocity and position respectively to achieve the optimization goal of minimum mean square error. Finally, the optimized neural network model is constructed using the position information of the optimal particle to complete the establishment of the IPSO-LSTM model.

[0053] Application examples of the prediction system and method of the present invention: The measured historical data of a photovoltaic power station in the past year of 2023 are selected, including the actual power generation data of the power station and meteorological data such as irradiance, temperature, wind speed, humidity and wind direction. The time scale of the data is 5 minutes per item.

[0054] First, the original measured data is preprocessed to remove outliers, missing values ​​or erroneous values ​​in the data. Based on the statistical characteristics of the photovoltaic irradiance sequence, the K-means++ clustering algorithm is used to perform generalized weather classification on the multiple types of weather contained in the power station.

[0055] Under different weather types, affected by irradiance and temperature, the power output characteristics of photovoltaic arrays vary significantly. By modeling each weather type condition separately, the model can capture the power output characteristics under specific weather types and reduce prediction errors. Therefore, after the weather types are broadly divided, the models under different generalized weather types are modeled and trained separately.

[0056] Based on the weather forecast of the day to be predicted, find 10 adjacent historical days of the same generalized weather type in the historical data, and then select the 5-day power data series with the smallest Euclidean distance of temperature as the model input. d i is the Euclidean distance between the temperature of the historical day and the day to be predicted, and the formula is: (14) Where: t 1 , t 2 are the maximum and minimum temperatures of the day to be predicted; t i1 , t i2 is the first i The highest and lowest temperatures in history.

[0057] Build a power prediction model based on generalized weather types. When building the IPSO-LSTM model under different generalized weather types, it is necessary to set the appropriate number of hidden layers and the number of neurons in each layer.

[0058] The multi-model power prediction system based on the generalized weather type classification proposed in this application example includes three prediction models IPSO-LSTM1, IPSO-LSTM2 and IPSO-LSTM3. The number of layers is set to 2, and the number of neurons is (128, 96), (256, 128), (256, 128) respectively. The improved adaptive noise complete set empirical mode decomposition is performed on the power generation data of the selected historical prediction day. The decomposed modes are used as model input. After normalization, the model output variable is the photovoltaic power generation power of the day to be predicted.

[0059] At the same time, under different generalized weather types, the prediction models based on back propagation neural network (BPNN) and kernel extreme learning machine (KELM) were compared, and the prediction accuracy was evaluated using the root mean square error (RMSE). The values ​​are shown in Table 2: Table 2 RMSE calculation results of different prediction models

[0060] It can be seen that under generalized weather types Ⅰ and Ⅱ, the RMSE of the IPSO-LSTM prediction model established by the present invention is smaller than that of the other two network models. Under generalized weather type Ⅲ, the RMSE of the IPSO-LSTM prediction model is smaller than that of the BPNN model and is close to that of the KELM model. This indicates that under different generalized weather types, the IPSO-LSTM prediction model of the present invention has better photovoltaic short-term power prediction accuracy than BPNN and KLEM.

[0061] The present invention also provides an electronic device, Figure 4 A schematic diagram of the structure of an electronic device provided by an embodiment of the present invention, such as Figure 4 As shown, the electronic device may include: a processor, a communications interface, a memory, and a communications bus, wherein the processor, the communications interface, and the memory communicate with each other via the communications bus. The processor may call the logic instructions in the memory, for example, to execute the following method: S1. Collect historical data on environmental factors of photovoltaic power generation, select influencing factors, and make a broad classification of weather types based on the influencing factors; S2. Collect historical data of photovoltaic power generation and normalize the historical power generation data; S3. Based on the broad classification of weather types, IPSO-LSTM models of different weather types are constructed and trained; S4. According to the weather type of the forecast day, select the IPSO-LSTM model of the corresponding weather type; S5. Selecting the corresponding normalized historical power generation data according to the weather type of the forecast day; S6. Based on the selected weather type IPSO-LSTM model, the selected normalized power generation data is used to predict the daily power generation power.

[0062] In addition, the logic instructions in the above-mentioned memory can be implemented in the form of software functional units and can be stored in a computer-readable storage medium when sold or used as an independent product. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0063] An embodiment of the present invention further provides a non-transitory computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the method provided in each of the above embodiments is implemented, for example, including: S1. Collect historical data on environmental factors of photovoltaic power generation, select influencing factors, and make a broad classification of weather types based on the influencing factors; S2. Collect historical data of photovoltaic power generation and normalize the historical power generation data; S3. Based on the broad classification of weather types, IPSO-LSTM models of different weather types are constructed and trained; S4. According to the weather type of the forecast day, select the IPSO-LSTM model of the corresponding weather type; S5. Selecting the corresponding normalized historical power generation data according to the weather type of the forecast day; S6. Based on the selected weather type IPSO-LSTM model, the selected normalized power generation data is used to predict the daily power generation power.

[0064] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0065] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0066] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A distributed photovoltaic power generation intelligent prediction system, characterized in that: include: The weather classification module broadly divides weather types based on historical environmental factor data; The data processing module performs data normalization based on the historical photovoltaic power generation data to reduce noise interference; The IPSO-LSTM module uses the trained weather type IPSO-LSTM model to perform intelligent prediction of distributed photovoltaic power generation based on the weather type of the forecast day and the normalized power generation data.

2. A prediction method based on the system of claim 1, characterized in that: Includes steps: S1. Collect historical data on environmental factors of photovoltaic power generation, select influencing factors, and make a broad classification of weather types based on the influencing factors; S2. Collect historical data of photovoltaic power generation and normalize the historical power generation data; S3. Based on the broad classification of weather types, IPSO-LSTM models of different weather types are constructed and trained; S4. According to the weather type of the forecast day, select the IPSO-LSTM model of the corresponding weather type; S5. Selecting the corresponding normalized historical power generation data according to the weather type of the forecast day; S6. Based on the selected weather type IPSO-LSTM model, the selected normalized power generation data is used to predict the daily power generation power.

3. The prediction method according to claim 2, characterized in that: When selecting influencing factors in S1, the Pearson correlation coefficient is used, and the formula is: (1) in, y is the photovoltaic power generation power; α are the relevant meteorological variable factors; t is the Pearson correlation coefficient.

4. The prediction method according to claim 3, characterized in that: When the weather types are broadly divided according to the influencing factors in S1, the generalized weather classification is performed based on K-means++ clustering, and the steps include: S11. Select the required number of cluster centers. The formula is: (2) (3) in, X For the data set, t is the number of cluster centers currently selected, μ i For the i The cluster center, D(x) is the distance between each point in the data set and the nearest point that has been selected as the cluster center; P(x) The distance of all points x is calculated D(x) After that, each point x The probability of being selected as the next cluster center; S12. Assign each point to the nearest cluster C j ; S13, according to each cluster C j The cluster center is updated with the mean value until the cluster center no longer changes or the change is less than the preset threshold; (4) Among them, | C j | for cluster C j The number of samples in .

5. The prediction method according to claim 2, characterized in that: When normalizing the historical power generation data in S2, the ICEEMDAN algorithm is used to perform multi-layer modal decomposition of the historical power generation data to obtain different modal components and residual components in each layer, thereby capturing the details and features of the historical power generation data and reducing noise interference.

6. The prediction method according to claim 5, characterized in that: The state update formula of the LSTM model in the IPSO-LSTM model is: （5） in, is the sigmoid function, tanh() is the hyperbolic tangent function, i t is the input gate, f t For the Gate of Forgetfulness, o t is the output gate, x t ，h t ，c t Respectively represent the input, output and cell state of LSTM at the current time step t, x t-1 ，h t-1 ，c t-1 Respectively represent the input, output and cell state of LSTM at time step t-1, W xi 、W hi 、W ci 、b i 、W xf 、W hf 、W cf 、b f 、W xc 、W hc 、b c 、W xo 、 W ho 、W co and b o are model parameters.

7. The prediction method according to claim 6, characterized in that: The network parameters of the LSTM model are optimized using the IPSO algorithm to construct an IPSO-LSTM model.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the steps of the testing method according to any one of claims 2 to 7 are implemented.

9. A non-transitory 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 testing method according to any one of claims 2 to 7 are implemented.

Images