A photovoltaic power generation amount prediction method and system

Abstract

The application discloses a photovoltaic power generation capacity prediction method and system, and relates to the photovoltaic power generation field. The method comprises the following steps: acquiring power curve data of a single user, pre-processing the power curve data, inputting the pre-processed power curve data into a photovoltaic power generation capacity prediction model, and predicting the power generation capacity of a whole area; and the training method of the photovoltaic power generation capacity prediction model comprises the following steps: collecting photovoltaic data sets of photovoltaic users in the same area, pre-processing the data sets; quantifying the influence of weather factors on photovoltaic power generation efficiency by using photovoltaic power curve data in the morning every day, combining photovoltaic power generation capacity data in a whole day, and constructing multiple regression models; comparing to obtain an optimal model; replacing photovoltaic data sets of photovoltaic users in the same area in a whole day with photovoltaic data sets of the whole area in a whole day, re-training the optimal model, and finally obtaining a photovoltaic power generation capacity prediction model. The method can predict the photovoltaic power generation capacity of the whole area by using the power curve data of a single user.

Description

Technical Field

[0001] This invention relates to the field of photovoltaic power generation, and proposes a method and system for predicting photovoltaic power generation. Background Technology

[0002] With the widespread application of distributed photovoltaic (PV) power generation technology, accurate forecasting of PV power generation has become a crucial aspect of grid operation and dispatch. Traditional PV power generation forecasting methods largely rely on meteorological data, but these methods often suffer from poor real-time performance and high data acquisition costs.

[0003] Chinese patent application CN117713188A discloses a data-physics jointly driven short-term photovoltaic (PV) output prediction method and system. This method constructs a physics-driven PV output prediction model by fitting weather information and its corresponding historical PV output data. This model can extract short-term changes in the relationship between weather factors and PV output. Then, a long-short-term time series network optimized by the whale optimization algorithm is used to train similar daily PV output data to obtain a data-driven PV output prediction model. This model can extract the long-term trend between weather factors and PV output. Finally, the Stacking ensemble learning framework is used to generalize the PV output prediction results of the two models, ensuring good prediction performance under both typical and transitional weather conditions. However, this method requires a large amount of data and cannot predict the entire day's power generation using morning PV power curve data. Summary of the Invention

[0004] To address the problems existing in the prior art, this invention proposes a photovoltaic power generation prediction method and system. This method can predict the photovoltaic power generation of the entire distribution area using the power curve data of a single user.

[0005] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows:

[0006] In a first aspect, the present invention provides a method for predicting photovoltaic power generation, comprising:

[0007] The power curve data of a single user is acquired, the power curve data is preprocessed, and the preprocessed power curve data is input into a photovoltaic power generation prediction model to predict the power generation of the entire distribution area; the training method of the photovoltaic power generation prediction model includes:

[0008] Collect photovoltaic datasets from photovoltaic users within the same distribution area, and preprocess the datasets.

[0009] The processed dataset is divided into a training set and a test set;

[0010] By using the photovoltaic power curve data every morning, the impact of weather factors on photovoltaic power generation efficiency is quantified, and various regression models are constructed by combining the photovoltaic power generation data throughout the day.

[0011] Multiple regression models were trained using the training set, and the root mean square error and mean absolute percentage error were calculated using the test set to evaluate the performance of the multiple regression models. The optimal model was then obtained by comparison.

[0012] Replace the photovoltaic dataset of all photovoltaic users in the same distribution area with the photovoltaic dataset of the entire distribution area for the whole day, retrain the optimal model, and finally obtain the photovoltaic power generation prediction model.

[0013] As a further improvement of the present invention, the step of collecting photovoltaic data sets from photovoltaic users within the same distribution area and performing data preprocessing on the dataset includes:

[0014] The power curve data and daily power generation data of photovoltaic users at 96 points each day are used as the dataset; the dataset is preprocessed, and the time and day length are adjusted; the data for all dates are adjusted for time and day length; after time adjustment, the corresponding photovoltaic power is obtained for each day; after day length adjustment, the adjusted power generation is obtained for each day.

[0015] As a further improvement of the present invention, the time adjustment includes: adjusting the sampling time point T of the photovoltaic power curve data. i Let i = 1, ..., 96, and let S be the sunrise time on day j. j Adjustments were made, and the sunrise time S1 of day 1 was selected as the baseline. The sampling time of the i-th sampling point on day j was:

[0016]

[0017] Adjusted sampling time T j i Photovoltaic power value P j i Based on the nearest sampling time point T k and T k+1 T k <T j i ≤T k+1 The corresponding photovoltaic power value P k and P k+1 It is calculated using linear interpolation, specifically as follows:

[0018] .

[0019] As a further improvement of the present invention, the day length adjustment includes: the power generation Q on day j. j Based on the length of daytime Lj The daytime length L1 of day 1 is selected as the baseline, and the standardized power generation W is obtained by standardization. j The power generation Q j Multiply by the base day length L1 and the current day length L j The ratio:

[0020] .

[0021] As a further improvement to the present invention, the construction of multiple regression models includes:

[0022] The first method is the linear regression model: A linear regression model is used to establish a linear equation describing the relationship between the morning photovoltaic power curve and the total photovoltaic power generation throughout the day.

[0023]

[0024] Where y represents the total photovoltaic power generation for the whole day, x1, x2, ..., x n This represents the photovoltaic power generation at different times in the morning, where β0 is the intercept, and β1, β2, ..., β n Here, ϵ represents the regression coefficients and the error term.

[0025] The second method is the ridge regression model: This model incorporates an L2 regularization term into the estimation of the regression coefficients.

[0026]

[0027] Where λ is the regularization parameter;

[0028] The third type is the LASSO regression model: The LASSO regression model incorporates an L1 regularization term.

[0029]

[0030] Where λ is the regularization parameter.

[0031] As a further improvement of the present invention, the root mean square error is a measure of the standard deviation between the model's predicted value and the actual value, specifically:

[0032]

[0033] In the formula, y i This represents the actual photovoltaic power generation on day i. i Let n represent the predicted photovoltaic power generation on day i, and n represent the total number of days.

[0034] As a further improvement of the present invention, the mean absolute percentage error measures the relative value of the prediction error, specifically as follows:

[0035]

[0036] In the formula, RMSE The root mean square error, y i Indicates the first i Actual power generation of Tianguang Solar ŷ i This represents the predicted photovoltaic power generation on day i. n Indicates the total number of days in the forecast; MAPE The mean absolute percentage error,

[0037] In a second aspect, the present invention provides a photovoltaic power generation prediction system, comprising:

[0038] The acquisition module is used to acquire power curve data for a single user.

[0039] The preprocessing module is used to preprocess the power curve data.

[0040] The prediction module is used to input the preprocessed power curve data into the photovoltaic power generation prediction model to predict the power generation of the entire area.

[0041] The training method for the photovoltaic power generation prediction model includes:

[0042] Collect photovoltaic datasets from photovoltaic users within the same distribution area, and preprocess the datasets.

[0043] The processed dataset is divided into a training set and a test set;

[0044] By using the photovoltaic power curve data every morning, the impact of weather factors on photovoltaic power generation efficiency is quantified, and various regression models are constructed by combining the photovoltaic power generation data throughout the day.

[0045] Multiple regression models were trained using the training set, and the root mean square error and mean absolute percentage error were calculated using the test set to evaluate the performance of the multiple regression models. The optimal model was then obtained by comparison.

[0046] Replace the photovoltaic dataset of all photovoltaic users in the same distribution area with the photovoltaic dataset of the entire distribution area for the whole day, retrain the optimal model, and finally obtain the photovoltaic power generation prediction model.

[0047] Thirdly, the present invention provides an electronic device including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the photovoltaic power generation prediction method.

[0048] Fourthly, the present invention provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the photovoltaic power generation prediction method.

[0049] Fifthly, the present invention provides a computer program product, the computer program product including computer instructions, the computer instructions instructing a computer to execute the photovoltaic power generation prediction method.

[0050] The advantages of this invention over the prior art are as follows:

[0051] This invention achieves effective prediction of daily photovoltaic (PV) power generation by establishing linear regression, ridge regression, and LASSO regression models. It also proposes a method for predicting the power generation of the entire distribution area using power curve data from a single user, which has significant application value in power grid dispatching. By establishing regression models and applying morning power data from PV users, the impact of weather factors on PV power generation efficiency is quantified, effectively predicting the daily PV power generation. Ridge regression and LASSO regression models are more robust in data processing than traditional linear regression models. Predicting the power generation of the entire distribution area using power curve data from a single user simplifies the calculation process and improves the efficiency and accuracy of prediction. Attached Figure Description

[0052] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following description is provided with accompanying drawings of the relevant technical solutions in the embodiments of the present invention or the prior art. It should be understood that the accompanying drawings described below are only for the purpose of clearly illustrating some embodiments of the technical solutions of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without any creative effort.

[0053] Figure 1 Here is a flowchart of a photovoltaic power generation prediction method provided by the present invention;

[0054] Figure 2 This is a flowchart illustrating the prediction process of an embodiment of the present invention;

[0055] Figure 3 The actual power generation of a photovoltaic user, the power generation fitted by the model, and the power generation curve predicted by the model;

[0056] Figure 4 This is a graph showing the power curve of a photovoltaic user and the total photovoltaic power curve of the corresponding transformer area.

[0057] Figure 5 The actual total power generation of the photovoltaic power generation in the distribution area, the model fitting total power generation, and the model prediction total power generation curve are shown.

[0058] Figure 6A photovoltaic power generation prediction device provided by the present invention;

[0059] Figure 7 This is a schematic diagram of an electronic device provided by the present invention. Detailed Implementation

[0060] The embodiments of the present invention are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention. The step numbers in the following embodiments are set only for ease of explanation, and there is no limitation on the order between the steps. The execution order of each step in the embodiments can be adaptively adjusted according to the understanding of those skilled in the art.

[0061] In the description of this invention, unless otherwise explicitly defined, terms such as "set up," "install," and "connect" should be interpreted broadly, and those skilled in the art can reasonably determine the specific meaning of the above terms in this invention in conjunction with the specific content of the technical solution.

[0062] Explanation of related terms

[0063] A linear regression model is a mathematical model that describes the linear relationship between two variables. It is used to describe the linear relationship between a dependent variable and one or more independent variables.

[0064] Ridge regression (Tikhonov regularization) is a biased estimation regression method specifically designed for collinear data analysis; it is essentially a modified least squares estimation method.

[0065] LASSO regression (Least Absolute Shrinkage and Selection Operator) is a regression analysis method that compresses variable coefficients by introducing L1 regularization.

[0066] This invention proposes a method for predicting the total photovoltaic (PV) power generation for the entire day based on morning PV power curve data, and it can predict the PV power generation of the entire distribution area using the power curve data of a single user. The purpose of this invention is to address the problem of inaccurate PV power generation prediction in existing technologies, particularly the shortcomings of traditional methods in terms of real-time performance and data acquisition costs.

[0067] The first objective of this invention is to provide a method for predicting photovoltaic power generation, comprising:

[0068] S100, acquires power curve data for a single user;

[0069] S200 performs data preprocessing on the power curve data;

[0070] S300, the preprocessed power curve data is input into the photovoltaic power generation prediction model to predict the power generation of the entire distribution area; the training method of the photovoltaic power generation prediction model includes:

[0071] S1, collect photovoltaic data sets from photovoltaic users within the same distribution area, and preprocess the data sets;

[0072] S2, divides the processed dataset into a training set and a test set;

[0073] S3 uses the photovoltaic power curve data from each morning to quantify the impact of weather factors on photovoltaic power generation efficiency, and combines the photovoltaic power generation data throughout the day to construct various regression models;

[0074] S4. Train multiple regression models using the training set, use the test set to calculate the root mean square error and mean absolute percentage error to evaluate the performance of multiple regression models, and compare to obtain the optimal model.

[0075] S5, replace the photovoltaic dataset of all photovoltaic users in the same area with the photovoltaic dataset of the entire area for the whole day, retrain the optimal model, and finally obtain the photovoltaic power generation prediction model.

[0076] The principle of this invention is as follows: It utilizes daily morning photovoltaic power curve data to quantify the impact of weather factors on photovoltaic power generation efficiency, and combines this with all-day photovoltaic power generation data to construct and optimize a predictive model. Specifically:

[0077] Power curve data is acquired from a single user's photovoltaic (PV) system, recording the system's power generation at different times. The data undergoes preprocessing, including outlier removal, missing value imputation, and smoothing, to ensure accuracy and consistency. Daily morning PV power curve data is used to analyze the impact of weather factors (such as sunlight intensity, temperature, and cloud thickness) on PV power generation efficiency. This can be achieved through feature extraction and pattern recognition of the power curves.

[0078] By combining daily photovoltaic (PV) power generation data, various regression models are constructed, such as linear regression, nonlinear regression, and machine learning algorithms (e.g., support vector machines, random forests). These models aim to learn the complex relationship between weather factors and PV power generation. The processed dataset is divided into training and testing sets. The training set is used to train the models, and the testing set is used to evaluate the models' performance.

[0079] Multiple regression models were trained using the training set, and the root mean square error (RMSE) and mean absolute percentage error (MAPE) were calculated using the test set to evaluate the models' predictive accuracy. By comparing the performance of different models, the optimal model was selected as the final photovoltaic power generation prediction model. The daily photovoltaic dataset for each photovoltaic user within the same distribution area was replaced with the daily photovoltaic dataset for the entire distribution area to more comprehensively reflect the photovoltaic power generation situation in the area. The optimal model was retrained to further optimize its predictive performance, ultimately resulting in a photovoltaic power generation prediction model applicable to the entire distribution area.

[0080] Specifically, this invention utilizes photovoltaic power curve data from each morning to quantify the impact of weather factors on photovoltaic power generation efficiency, and combines this with all-day photovoltaic power generation data to establish an effective prediction model; it predicts the power generation of the entire distribution area using the power curve data of a single user, thereby improving the accuracy and ease of calculation of power generation prediction for the entire distribution area.

[0081] Furthermore, this invention quantifies the impact of weather factors on photovoltaic power generation efficiency using photovoltaic power curve data from each morning, which helps improve the accuracy and robustness of the prediction model. By collecting power curve data from a single user and using a pre-trained prediction model to predict the power generation of the entire distribution area, the prediction process is simplified and the accuracy of the prediction is improved. By constructing multiple regression models and evaluating their performance, this invention can select the optimal prediction model to adapt to the photovoltaic power generation prediction needs under different scenarios. This method achieves accurate prediction of photovoltaic power generation through steps such as quantifying the impact of weather factors, predicting the power generation of the entire distribution area using single-user data, and constructing and selecting the optimal prediction model.

[0082] The method of the present invention will be described in detail below with reference to specific drawings and embodiments:

[0083] like Figure 2 As shown, this invention provides a method for predicting the total photovoltaic power generation for the entire day based on morning photovoltaic power curve data, and predicts the power generation of the entire distribution area using the power curve data of a single user. The training method for the photovoltaic power generation prediction model is implemented through the following steps:

[0084] S1. Data preprocessing: Collect photovoltaic datasets and perform data preprocessing on the datasets, including time adjustment and day length adjustment.

[0085] Time adjustment: Adjust the sampling time point T of the photovoltaic power curve data. i (i=1,...,96), based on the sunrise time S of day j. jAdjustments were made, selecting the sunrise time S1 of day 1 as the baseline. This was done to ensure that the solar altitude at the i-th sampling point on day j was relatively consistent with the solar altitude at the i-th sampling point on day 1, thus eliminating errors caused by differences in solar altitude. The sampling time at the i-th sampling point on day j was determined.

[0086]

[0087] Adjusted sampling time T j i Photovoltaic power value P j i The nearest sampling time point can be used as T. k and T k+1 (T) k <T j i ≤T k+1 The corresponding photovoltaic power value P k and P k+1 The result was obtained through linear interpolation:

[0088]

[0089] Day length adjustment: Power generation Q on day j j Based on the length of daytime L j The daytime length L1 of day 1 is selected as the baseline, and the standardized power generation W is obtained by standardization. j The upcoming power generation Q j Multiply by the base day length L1 and the current day length L j The ratio is used to eliminate the impact of day length differences on power generation, which facilitates accurate model training and prediction results.

[0090]

[0091] S2 Dataset Partitioning: The processed dataset is divided into a training set and a test set. The training set is used for model training, and the test set is used for model validation and performance evaluation.

[0092] S3. Model Establishment: This invention employs three regression models to predict a user's daily photovoltaic power generation:

[0093] Linear Regression Model: This model uses the most basic linear regression model to describe the relationship between the morning photovoltaic power curve and the total photovoltaic power generation throughout the day by establishing a linear equation. The model is as follows:

[0094]

[0095] Where y represents the total photovoltaic power generation for the whole day, x1, x2, ..., x nThis represents the photovoltaic power generation at different times in the morning, where β0 is the intercept, and β1, β2, ..., β n ϵ is the regression coefficient and ϵ is the error term.

[0096] Ridge Regression Model: To address the multicollinearity problem that may occur in linear regression, the ridge regression model is introduced. This model incorporates an L2 regularization term in the estimation of regression coefficients, making the model more robust. Its mathematical form is:

[0097]

[0098] Here, λ is the regularization parameter, which controls the complexity of the model.

[0099] LASSO Regression Model: Similar to the Ridge Regression Model, the LASSO regression model controls the magnitude of the regression coefficients by adding an L1 regularization term, thus achieving feature selection. Its mathematical form is:

[0100]

[0101] Where λ is the regularization parameter.

[0102] S4. Model Training and Testing: Train the three regression models, use the test set to calculate the root mean square error (RMSE) and mean absolute percentage error (MAPE) to evaluate the model performance, and compare to obtain the optimal model.

[0103] RMSE: Measures the standard deviation between the model's predicted value and the actual value. The formula is:

[0104]

[0105] MAPE: Measures the relative value of prediction error, and the formula is:

[0106]

[0107] In the formula, RMSE The root mean square error, y i Indicates the first i Actual power generation of Tianguang Solar ŷ i This represents the predicted photovoltaic power generation on day i. n Indicates the total number of days in the forecast; MAPE The mean absolute percentage error,

[0108] S5. Predicting total power generation across the entire region based on a single user.

[0109] Since photovoltaic users within the same distribution area have similar solar radiation environments, their power curves exhibit similarity. Based on this, this invention proposes a method for predicting the total power generation of the entire distribution area using the power curve data of a single user. This method involves repeating steps S3 and S4, replacing the user's total daily power generation with the total daily power generation of the entire distribution area, retraining the model, comparing the error results, and finally obtaining the optimal model. By extending the prediction results of a single user to the entire distribution area, the simplicity and efficiency of predicting the total power generation of the entire distribution area are improved.

[0110] Example

[0111] After preprocessing, the data was divided into training and testing sets. Linear regression, ridge regression, and LASSO regression models were used for training, and their performance was validated on the testing set to select the optimal model. Results show that the method of this invention has high accuracy in predicting daily power generation. The method for predicting the power generation of the entire distribution area using a single user's power curve demonstrates good practicality and accuracy.

[0112] The following specific embodiments illustrate the implementation of the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. A method for predicting power generation and consumption based on a regression model ( Figure 2 ), including the following steps:

[0113] Step 1: Select the daily power curve data and daily power generation data of photovoltaic users in a photovoltaic area of ​​Shunping County, Baoding City, Hebei Province from February 15 to March 15, 2024 as the dataset.

[0114] Furthermore, the dataset is preprocessed, and time and day length adjustments are made to the dataset.

[0115] Based on February 15, 2024, the sunrise time was 07:11:12, and the day length was 10 hours, 44 minutes, and 46 seconds. Taking February 16, 2024 as an example, the sunrise time was 07:09:59, and the day length was 10 hours, 47 minutes, and 07 seconds, which was 1 minute and 13 seconds earlier than the sunrise time on February 15, 2024, and the day length was 2 minutes and 21 seconds longer. Since the photovoltaic power at point 96 is recorded in 15-minute increments, the solar altitude at 8:00 on February 15, 2024, should be similar to the solar altitude at 7:58:47 on February 16, 2024. The photovoltaic power at 7:58:47 on February 16, 2024, can be calculated from the photovoltaic power at 7:45 and 8:00 on the same day using linear interpolation.

[0116] The power generation on February 16, 2024, was adjusted by multiplying the daytime length by an adjustment ratio, which is the ratio of the daytime length on February 15, 2024 (the baseline) to the daytime length on February 16, 2024. Similarly, time and daytime length adjustments were made to the data for all dates.

[0117] After time adjustment, five time points were selected on February 15, 2024: 8:00, 8:15, 8:30, 8:45, and 9:00 to obtain the corresponding photovoltaic power for each day after time adjustment. After adjusting for day length, the adjusted power generation for each day was obtained.

[0118] Step 2: After preprocessing, the data from February 15 to March 6, 2024, is used as the training set, and the data from March 7 to March 15, 2024, is used as the test set.

[0119] Step 3: Based on the dataset from Step 2, use the morning user photovoltaic power curve data as the independent variable in the regression model, and the user's total daily power generation as the dependent variable. By selecting different combinations of independent variables, input the data into the constructed linear regression, ridge regression, and LASSO regression models respectively to obtain the predicted power generation during the test set time period.

[0120] Step 4: Compare the prediction errors of the three regression models and output the result of the optimal model.

[0121] Step 5: According to Figure 3 In this context, the correlation between the power generation of photovoltaic users and the power generation of the distribution area can be used to replace the user's daily power generation forecast with the total daily power generation of the distribution area. According to... Figure 4 The correlation between the power generation of photovoltaic users and the power generation of the photovoltaic system in the distribution area is analyzed to predict the power generation of photovoltaic users. Steps 3 and 4 are repeated, replacing the dependent variable of the user's total daily power generation with the total daily power generation of the distribution area, to predict the power generation of the distribution area. Finally, the results of the optimal model are output, see [link to model]. Figure 5 .

[0122] This embodiment describes a method for preprocessing photovoltaic power curves and photovoltaic power generation to make the data usable for model training. Preprocessing the photovoltaic power curves allows for time adjustment, ensuring that different dates are at similar solar altitudes, thus eliminating the influence of sunrise / sunset times. Photovoltaic power output reflects the impact of weather factors. Preprocessing photovoltaic power generation allows for day length adjustment, preventing the impact of varying day lengths on power generation on different dates, thereby affecting model training and prediction.

[0123] Furthermore, a photovoltaic power generation prediction method based on regression models is provided. Linear regression, ridge regression, and LASSO regression models are established, and the accuracy of the whole day's photovoltaic power generation prediction is improved by utilizing morning photovoltaic power curve data. Based on the power curve data of a single user, the power generation of the entire distribution area is predicted by utilizing the correlation of photovoltaic users' power generation in the same distribution area.

[0124] In summary, this invention provides a method for predicting all-day photovoltaic power generation based on morning photovoltaic power curve data and a regression model. This method offers high prediction accuracy and can predict the power generation of the entire distribution area using data from a single user. Therefore, by establishing a regression model and using morning photovoltaic power as the independent variable to predict all-day photovoltaic power generation, this invention avoids the lack of real-time capability in weather forecasts and improves the accuracy of photovoltaic power generation prediction.

[0125] like Figure 6 As shown, a third objective of this invention is to provide a photovoltaic power generation prediction system, comprising:

[0126] Acquisition module 100 is used to acquire power curve data for a single user;

[0127] Preprocessing module 200 is used to preprocess the power curve data;

[0128] The prediction module 300 is used to input the preprocessed power curve data into the photovoltaic power generation prediction model to predict the power generation of the entire area.

[0129] The training method for the photovoltaic power generation prediction model includes:

[0130] Collect photovoltaic datasets from photovoltaic users within the same distribution area, and preprocess the datasets.

[0131] The processed dataset is divided into a training set and a test set;

[0132] By using the photovoltaic power curve data every morning, the impact of weather factors on photovoltaic power generation efficiency is quantified, and various regression models are constructed by combining the photovoltaic power generation data throughout the day.

[0133] Multiple regression models were trained using the training set, and the root mean square error and mean absolute percentage error were calculated using the test set to evaluate the performance of the multiple regression models. The optimal model was then obtained by comparison.

[0134] Replace the photovoltaic dataset of all photovoltaic users in the same distribution area with the photovoltaic dataset of the entire distribution area for the whole day, retrain the optimal model, and finally obtain the photovoltaic power generation prediction model.

[0135] The photovoltaic power generation prediction system of the present invention is based on the photovoltaic power generation prediction method.

[0136] like Figure 7 As shown, a third objective of this invention is to provide an electronic device, including a memory 701, a processor 702, and a computer program stored in the memory 701 and executable on the processor. When the processor executes the computer program, it implements the aforementioned photovoltaic power generation prediction method. The device also includes a communication interface 703 and a bus 704.

[0137] A fourth objective of this invention is to provide a computer-readable storage medium storing a computer program that, when executed by a processor, implements the aforementioned photovoltaic power generation prediction method.

[0138] A fifth objective of this invention is to provide a computer program product comprising computer instructions that instruct a computer to execute the aforementioned photovoltaic power generation prediction method.

[0139] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0140] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0141] This invention may take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this invention may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, readable storage media, optical storage, etc.) containing computer-usable program code.

[0142] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0143] Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort should fall within the scope of protection of the present invention.

[0144] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A method for predicting photovoltaic power generation, characterized in that, include: The power curve data of a single user is obtained, the power curve data is preprocessed, and the preprocessed power curve data is input into the photovoltaic power generation prediction model to predict the power generation of the entire area. The training method for the photovoltaic power generation prediction model includes: Collect photovoltaic datasets from photovoltaic users within the same distribution area, and preprocess the datasets. The processed dataset is divided into a training set and a test set; By using the photovoltaic power curve data every morning, the impact of weather factors on photovoltaic power generation efficiency is quantified, and various regression models are constructed by combining the photovoltaic power generation data throughout the day. Multiple regression models were trained using the training set, and the root mean square error and mean absolute percentage error were calculated using the test set to evaluate the performance of the multiple regression models. The optimal model was then obtained by comparison. Replace the photovoltaic dataset of all photovoltaic users in the same area with the photovoltaic dataset of the entire area for the whole day, retrain the optimal model, and finally obtain the photovoltaic power generation prediction model. The process of collecting photovoltaic data sets from photovoltaic users within the same distribution area and preprocessing the dataset includes: The power curve data and daily power generation data of photovoltaic users at 96 points each day are used as the dataset; the dataset is preprocessed, and time and day length adjustments are performed on the dataset; time and day length adjustments are performed on the data for all dates; after time adjustment, the corresponding photovoltaic power after time adjustment is obtained; after day length adjustment, the adjusted power generation for each day is obtained. The time adjustment comprises: adjusting the sampling time point T i of the photovoltaic power curve data, i=1,...,96, according to the sunrise time S j of the jth day, taking the sunrise time S1 of the 1st day as the reference, and the sampling time of the ith sampling point of the jth day is: adjusted sampling time T j i photovoltaic power value P j i , according to the nearest sampling time point for T k and T k+1 , T k <T j i ≤T k+1 , the corresponding photovoltaic power value P k and P k+1 , are calculated by linear interpolation method, specifically: 。 2. The photovoltaic power generation prediction method according to claim 1, characterized in that, The day length adjustment includes: the power generation Q on day j. j Based on the length of daytime L j The daytime length L1 of day 1 is selected as the baseline, and the standardized power generation W is obtained by standardization. j The power generation Q j Multiply by the base day length L1 and the current day length L j The ratio: 。 3. The photovoltaic power generation prediction method according to claim 1, characterized in that, The construction of multiple regression models includes: The first method is the linear regression model: A linear regression model is used to establish a linear equation describing the relationship between the morning photovoltaic power curve and the total photovoltaic power generation throughout the day. Where y represents the total photovoltaic power generation for the whole day, x1, x2, ..., x n This represents the photovoltaic power generation at different times in the morning, where β0 is the intercept, and β1, β2, ..., β n For regression coefficients, This is the error term; The second method is the ridge regression model: This model incorporates an L2 regularization term into the estimation of the regression coefficients. Where λ is the regularization parameter; The third type is the LASSO regression model: The LASSO regression model incorporates an L1 regularization term. Where λ is the regularization parameter.

4. The photovoltaic power generation prediction method according to claim 1, characterized in that, The root mean square error is a measure of the standard deviation between the model's predicted value and the actual value, specifically: In the formula, RMSE The root mean square error, y j Indicates the first j Actual power generation of Tianguang Solar j This represents the predicted photovoltaic power generation on day j. n Indicates the total number of days in the forecast; The mean absolute percentage error measures the relative value of the prediction error, specifically: In the formula, MAPE The mean absolute percentage error.

5. A photovoltaic power generation prediction system, implementing the photovoltaic power generation prediction method according to any one of claims 1-4, characterized in that, include: The acquisition module is used to acquire power curve data for a single user. The preprocessing module is used to preprocess the power curve data. The prediction module is used to input the preprocessed power curve data into the photovoltaic power generation prediction model to predict the power generation of the entire area. The training method for the photovoltaic power generation prediction model includes: Collect photovoltaic datasets from photovoltaic users within the same distribution area, and preprocess the datasets. The processed dataset is divided into a training set and a test set; By using the photovoltaic power curve data every morning, the impact of weather factors on photovoltaic power generation efficiency is quantified, and various regression models are constructed by combining the photovoltaic power generation data throughout the day. Multiple regression models were trained using the training set, and the root mean square error and mean absolute percentage error were calculated using the test set to evaluate the performance of the multiple regression models. The optimal model was then obtained by comparison. Replace the photovoltaic dataset of all photovoltaic users in the same distribution area with the photovoltaic dataset of the entire distribution area for the whole day, retrain the optimal model, and finally obtain the photovoltaic power generation prediction model.

6. An electronic device, characterized in that, The method includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the computer program, implements the photovoltaic power generation prediction method according to any one of claims 1-4.

7. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the photovoltaic power generation prediction method according to any one of claims 1-4.

8. A computer program product, said computer program product comprising computer instructions, characterized in that, The computer instructions instruct the computer to execute the photovoltaic power generation prediction method according to any one of claims 1-4.

Images