A photovoltaic power generation prediction method, system, device and storage medium

Abstract

The application discloses a kind of photovoltaic power generation prediction method, system, device and storage medium, method includes: according to the historical load data of photovoltaic power station, weather factor correlation analysis is carried out combining meteorological factor historical data, determine target meteorological factor;According to the historical load data of photovoltaic power station, carry out load sequence correlation analysis, determine the target specification of load data;Build prediction neural network;Get the historical load data of photovoltaic power station target specification, and the historical data of target meteorological factor associated with the historical load data of target specification;According to the historical load data of target specification and the historical data of target meteorological factor, model training is carried out to prediction neural network, and prediction model is obtained;According to the data of target meteorological factor of day to be predicted, photovoltaic power generation is predicted by prediction model.The present application can improve prediction accuracy by meteorological factor data for photovoltaic power generation prediction, and can be widely applied in photovoltaic power generation prediction technical field.

Description

Technical Field

[0001] This invention relates to the field of photovoltaic power generation prediction technology, and in particular to a photovoltaic power generation prediction method, system, device and storage medium. Background Technology

[0002] Solar energy, as a renewable and clean energy source, is gaining increasing popularity and attention. Accurate forecasting of photovoltaic (PV) output is crucial for achieving local consumption of distributed PV power and enhancing the grid's capacity to accommodate PV power. It also supports optimized planning and design of PV power sources, as well as optimized scheduling and management of the grid. However, because PV power generation is closely related to weather factors (solar radiation, temperature, etc.), and the variability of these factors causes random fluctuations in PV power generation, accurate forecasting of PV output is challenging. Summary of the Invention

[0003] In view of this, embodiments of the present invention provide a photovoltaic power generation prediction method, system, device, and storage medium.

[0004] On one hand, embodiments of the present invention provide a photovoltaic power generation prediction method, including:

[0005] Based on the historical load data of photovoltaic power stations, and combined with historical meteorological data, a correlation analysis of weather factors is conducted to determine the target meteorological factors.

[0006] Based on the historical load data of the photovoltaic power station, load sequence correlation analysis is performed to determine the target specifications for the load data;

[0007] Construct a predictive neural network;

[0008] Acquire historical load data of the photovoltaic power station for the target specifications, and historical data of the target meteorological factors associated with the historical load data of the target specifications;

[0009] Based on the historical load data of the target specification and the historical data of the target meteorological factors, the prediction neural network is trained to obtain the prediction model;

[0010] Photovoltaic power generation is predicted using the prediction model based on the target meteorological factors data for the predicted date.

[0011] Optionally, it also includes:

[0012] The historical load data of the target specification is preprocessed.

[0013] Optionally, the data preprocessing of the historical load data for the target specification includes:

[0014] Missing values ​​in the historical load data of the target specification are filled in using a single-point completion method.

[0015] And / or, use a multi-point completion method to fill in the missing values ​​of the historical load data of the target specification.

[0016] Optionally, the data preprocessing of the historical load data for the target specification includes:

[0017] Based on a preset threshold, determine the glitch data of the historical load data of the target specification;

[0018] The burr data is replaced according to a preset processing method.

[0019] Optionally, the step of determining the target meteorological factor by performing a correlation analysis of weather factors based on historical load data of the photovoltaic power station and historical meteorological factor data includes:

[0020] Based on the historical load data of photovoltaic power stations and combined with historical meteorological data, the correlation analysis of each meteorological factor is carried out sequentially using the correlation coefficient calculation formula to determine the target meteorological factors affecting the load of photovoltaic power stations.

[0021] The formula for calculating the correlation coefficient is as follows:

[0022]

[0023] In the formula, p xy The correlation coefficient is represented by n, where n represents the total amount of data, and x represents the total number of data points. i This represents the i-th historical load data. y represents the average of historical load data. i This represents the historical data of the i-th meteorological factor. This represents the average value of historical data for meteorological factors.

[0024] Optionally, the step of performing load sequence correlation analysis based on historical load data of the photovoltaic power station to determine the target specifications of the load data includes:

[0025] Based on the historical load data of photovoltaic power plants, nearest neighbor correlation analysis and periodic correlation analysis of load sequences are performed to determine the target specifications of load data;

[0026] The periodic correlation analysis includes daily periodic correlation analysis, weekly periodic analysis, and annual periodic analysis.

[0027] Optionally, it also includes:

[0028] Based on the load forecast and actual load values ​​from the prediction model, the model parameters of the prediction model are corrected using an error function.

[0029] Secondly, embodiments of the present invention provide a photovoltaic power generation prediction system, comprising:

[0030] The first module is used to perform correlation analysis of weather factors based on historical load data of photovoltaic power plants and historical meteorological factor data to determine target meteorological factors.

[0031] The second module is used to perform load sequence correlation analysis based on the historical load data of the photovoltaic power station to determine the target specifications of the load data;

[0032] The third module is used to build the predictive neural network;

[0033] The fourth module is used to acquire historical load data of the photovoltaic power station for the target specifications, and historical data of the target meteorological factors associated with the historical load data of the target specifications;

[0034] The fifth module is used to train the prediction neural network based on the historical load data of the target specification and the historical data of the target meteorological factors to obtain a prediction model;

[0035] The sixth module is used to predict photovoltaic power generation based on the target meteorological factors data for the date to be predicted, using the prediction model.

[0036] On the other hand, embodiments of the present invention provide a photovoltaic power generation prediction device, including a processor and a memory;

[0037] The memory is used to store programs;

[0038] The processor executes the program to implement the method described above.

[0039] On the other hand, embodiments of the present invention provide a computer-readable storage medium storing a program that is executed by a processor to implement the methods described above.

[0040] This invention also discloses a computer program product or computer program, which includes computer instructions stored in a computer-readable storage medium. A processor of a computer device can read the computer instructions from the computer-readable storage medium and execute the computer instructions, causing the computer device to perform the aforementioned method.

[0041] The embodiments of this invention first perform weather factor correlation analysis based on historical load data of photovoltaic power plants and historical meteorological factor data to determine target meteorological factors; then, based on historical load data of photovoltaic power plants, perform load sequence correlation analysis to determine the target specifications of the load data; construct a prediction neural network; acquire historical load data of the photovoltaic power plant at the target specifications, and historical data of the target meteorological factors associated with the historical load data of the target specifications; train the prediction neural network model based on the historical load data of the target specifications and the historical data of the target meteorological factors to obtain a prediction model; and then, based on the data of the target meteorological factors on the day to be predicted, perform photovoltaic power generation prediction through the prediction model. This invention determines the target meteorological factors affecting the load of photovoltaic power plants through weather factor correlation analysis, avoiding interference from irrelevant weather factors; subsequently, it determines the target specifications of the load data through load sequence correlation analysis, improving prediction accuracy based on the data correlation of the target specifications; simultaneously, by training the prediction model through the neural network based on the correlation between load data and meteorological factors, it is possible to achieve photovoltaic power generation prediction based on meteorological factor data. Attached Figure Description

[0042] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0043] Figure 1 This is a schematic diagram of the overall steps of the photovoltaic power generation prediction method provided in the embodiments of the present invention;

[0044] Figure 2 A schematic diagram of the overall steps of the photovoltaic power generation prediction method provided in the embodiments of the present invention;

[0045] Figure 3 A schematic diagram illustrating the correlation between wind speed, radiation value, and photovoltaic power generation power provided in an embodiment of the present invention;

[0046] Figure 4 This is a schematic diagram illustrating the correlation between power generation sequences and power generation series provided in an embodiment of the present invention;

[0047] Figure 5 A schematic diagram of a predictive learning training queue provided in an embodiment of the present invention;

[0048] Figure 6 A schematic diagram of a photovoltaic power generation prediction system provided in an embodiment of the present invention;

[0049] Figure 7This is a schematic diagram of a photovoltaic power generation prediction device provided in an embodiment of the present invention. Detailed Implementation

[0050] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0051] To address the problems existing in the prior art, this invention provides a photovoltaic power generation prediction method, such as... Figure 1 As shown, the method includes the following steps:

[0052] S100. Based on the historical load data of the photovoltaic power station and combined with the historical data of meteorological factors, conduct a correlation analysis of weather factors to determine the target meteorological factors.

[0053] It should be noted that, based on historical load data of the photovoltaic power station and combined with historical meteorological data, correlation analysis of each meteorological factor was conducted sequentially using the correlation coefficient calculation formula to determine the target meteorological factors affecting the load of the photovoltaic power station; the correlation coefficient calculation formula is as follows:

[0054]

[0055] In the formula, p xy The correlation coefficient is represented by n, where n represents the total amount of data, and x represents the total number of data points. i This represents the i-th historical load data. y represents the average of historical load data. i This represents the historical data of the i-th meteorological factor. This represents the average value of historical data for meteorological factors.

[0056] Specifically, based on the historical power generation data (i.e., historical load data) of the photovoltaic power station, eigenvector correlation analysis (i.e., weather factor correlation analysis) is performed with various meteorological factors, such as the correlation analysis between wind speed, radiation value, and photovoltaic power generation. The analysis results are as follows: Figure 2 As shown, it is easy to see that the power generation is significantly correlated with the radiation value, but not with wind speed. Besides being affected by solar irradiance, the output power of a photovoltaic system also varies considerably under different weather conditions. On sunny days, the photovoltaic power output is stable, basically following a normal distribution, with the maximum output power occurring around noon; on cloudy days, the photovoltaic power output exhibits greater randomness and fluctuation. As the outside temperature gradually rises, the conversion efficiency of the photovoltaic power generation system will gradually decrease, thus reducing the output power. Under the same weather conditions, the power output curves at different ambient temperatures show inconsistent heights.

[0057] Correlation analysis of weather factors refers to the study of the degree of correlation between any two variables. It is only meaningful to study the mutual influence between variables when there is a certain correlation and dependence between them. Correlation analysis is a method for calculating the degree of connection between variables. The formula for calculating the correlation coefficient is as follows:

[0058]

[0059] In the formula, p xy The correlation coefficient is represented by n, where n represents the total amount of data, and x represents the total number of data points. i This represents the i-th historical load data. This represents the average value of historical load data. y i This represents the historical data of the i-th meteorological factor. This represents the average value of historical data for meteorological factors. When |p xy When | = 1, it indicates that the load data and the corresponding meteorological factors are completely correlated; when | p xy When | = 0, it is unrelated; when | p xy | > 0.8 indicates extremely strong correlation; when |p xy A correlation coefficient of less than 0.2 indicates extremely weak correlation; other values ​​indicate moderate correlation. Based on the data correlation analysis, solar radiation, temperature, humidity, and wind speed were selected as target meteorological factors, and then the meteorological data of solar radiation, temperature, humidity, and wind speed were used as the photovoltaic prediction feature vector.

[0060] S200. Based on the historical load data of the photovoltaic power station, perform load sequence correlation analysis to determine the target specifications of the load data;

[0061] It should be noted that, based on the historical load data of the photovoltaic power station, nearest neighbor correlation analysis and periodic correlation analysis are performed on the load sequence to determine the target specifications of the load data; among which, the periodic correlation analysis includes daily periodic correlation analysis, weekly periodic analysis and annual periodic analysis.

[0062] Specifically, such as Figure 3 As shown, the correlation of photovoltaic power generation can be broadly categorized into two types: the correlation between power generation sequences (i.e., load sequences) and power generation sequences. The correlation types can be further subdivided into nearest-neighbor correlation and periodic correlation. Periodic correlation can be further subdivided into daily periodicity, weekly periodicity, annual periodicity, etc. The principle of sequence correlation analysis is similar to that of weather factor correlation analysis, and will not be elaborated upon here.

[0063] Neighbor-related:

[0064] Load correlation analysis involves examining two load sequences extracted from the same time point and the previous time point on multiple dates. This correlation analysis can demonstrate a high degree of correlation between two load sequences with adjacent time points. If the correlation is high, the load data from adjacent values ​​are considered as a contributing factor, influencing the predicted values.

[0065] Cycle-related:

[0066] Load is cyclically related, reflected in the correlation between daily curves of the same type of date. If daily periodicity is used as the metric, then two adjacent days are taken as the data; if weekly periodicity is used, then two days every other week are taken as the data, and so on. The advantage of using different periodicities lies in distinguishing between ordinary and special days. For example, daily periodic correlation is generally higher for weekdays, weekly periodic correlation is generally higher for weekends, and annual periodic correlation is generally higher for holidays. After calculation, using a highly correlated sequence as the benchmark for prediction can undoubtedly improve prediction accuracy very quickly.

[0067] S300, Construct a predictive neural network;

[0068] S400: Obtain historical load data of the photovoltaic power station target specifications, and historical data of the target meteorological factors associated with the historical load data of the target specifications;

[0069] S500. Based on the historical load data of the target specification and the historical data of the target meteorological factors, the prediction neural network is trained to obtain the prediction model.

[0070] Specifically, the artificial intelligence algorithms employed in this invention include KNN, Random Forest, Decision Tree, Bayesian algorithm, and Multi-Level Photonics (MLP). The artificial intelligence algorithms and prediction algorithms mentioned later refer to these algorithms. These artificial intelligence algorithms possess strong adaptive and self-learning capabilities. They can use historical load data and historical data of factors influencing the load (i.e., target meteorological factors) as input, iteratively calling KNN, Random Forest, Decision Tree, Bayesian algorithm, and Multi-Level Photonics (MLP) algorithms to predict the results of different algorithm models. Individual algorithm error analysis and combined prediction error analysis are then performed (the analysis method is explained in the supplementary section on the inverse variance method of combined prediction error). The model with the smallest error analysis result is selected as the prediction model. This allows for the training of a load value model without requiring explicit mathematical expressions. Load forecasting refers to using a set of mathematical methods to process the relationship between historical and future load values, taking into full account the influence of various factors, thereby estimating the power generation load value of a photovoltaic power station at a future time.

[0071] A prediction algorithm is a learning algorithm that requires a learned prediction model to make predictions. Training the prediction model requires training samples and validation samples. Training samples are used to train the model, while validation samples are used to check the model's state and convergence during training. Training and validation samples consist of multiple samples, each in the form: Y; X1 X2 X3 X4…Xn. Here, Y is the predicted value, and X represents the factors used to obtain the Y value; X can be correlated with the sample parameters. The time period from which data is used determines the number of training samples, which in turn determines the training time and accuracy. Therefore, a balance should be struck when selecting training samples to achieve the best results.

[0072] S600: Based on the target meteorological factor data for the day to be predicted, photovoltaic power generation is predicted using a prediction model.

[0073] Specifically, the learning and training queue is as follows: Figure 4 As shown, the prediction method (e.g., the preset specification is to select 20 days of samples):

[0074] (1) Study the load of the first 20 days and the meteorological factors that affect the load, and predict the load of the 21st day based on the real-time meteorological factors of the 21st day.

[0075] (2) Study the load from day 2 to day 21 and the meteorological factors that affect the load, predict the load on day 22, and so on.

[0076] In some embodiments, the method further includes: preprocessing historical load data for the target specification.

[0077] It should be noted that missing values ​​in the historical load data of the target specification are filled in using a single-point completion method; and / or, missing values ​​in the historical load data of the target specification are filled in using a multi-point completion method.

[0078] Specifically, handling missing data values:

[0079] ① Single-point completion method: also known as the averaging method, if neither the left nor right value of the missing point is empty, the average value is entered. This method will never fail to repair.

[0080] ② Multi-point completion method: also known as the similar-day method. If multiple consecutive data points are missing, or if the missing points are at both ends, making it impossible to use the left and right values ​​for averaging, then the load data at the corresponding time points of the first 3 similar days and the last 3 similar days are obtained, and the average value is entered. Generally, a 6-point average is taken. If two of the similar days also have missing data at that time point, then a 4-point average is taken. If all 6 similar days have missing data at that time point, then completion for that point fails, and the missing point will continue to exist in the database with a value of 0. Therefore, it is crucial to ensure that there are no consecutive missing data points for multiple days in the database; otherwise, completion will definitely fail.

[0081] It should also be noted that, based on a preset threshold, glitch data of historical load data for the target specification is determined; and the glitch data is replaced according to a preset processing method.

[0082] Specifically, data glitches handling:

[0083] A preset threshold (e.g., n times or 1 / n of the load at the previous and next time points) is used. When val > a preset positive threshold (i.e., the load at the current time point is greater than n times the load at the previous and next time points) or val < a preset negative threshold (i.e., the load at the current time point is less than 1 / n of the load at the previous and next time points), values ​​outside the range are processed by glitching. Here, val represents the value collected at the current time point. Processing methods include:

[0084] ① Use previous value: Replace the glitch data with the value from the previous time step.

[0085] ② Use the later value: Replace the glitch data with the value at the next later time.

[0086] ③ Use average value: Replace the spurious data with the average value of the values ​​before and after the time step.

[0087] In some embodiments, the method further includes: correcting the model parameters of the prediction model using an error function based on the predicted load value and the actual load value of the prediction model.

[0088] Specifically, such as Figure 5 As shown, y(t) is the output value at time t, which is also the predicted value at time t. It is obtained by the model at time t-1, the actual value x(t-1), and the parameters at time t. E(t) is the difference between y(t) and the actual value x(t). If E(t) = 0, the model parameters are not corrected. Otherwise, the model parameters should be corrected to track changes in the environment.

[0089] The formula for solving the error function E is as follows:

[0090]

[0091] In the formula, n represents the number of photovoltaic data points collected in a day; i represents the i-th point collected in a day; x i y represents the i-th actual value;i Let t represent the i-th predicted value; E represents the error value, which is E(t). The time t corresponds to the i-th point collected: for example, photovoltaic data is collected at one point every 15 minutes, and time t = i / 4 hours (i%4) * 15 minutes. If i = 96, then t is 0:00.

[0092] The combined forecasting error analysis employs the inverse variance method: the sum of squared forecasting errors is an indicator of forecasting accuracy. A larger sum of squared forecasting errors indicates lower forecasting accuracy for that particular forecasting model, thus reducing its importance in the combined forecasting. Importance is reflected in its weight w within the combined forecasting model. j The calculation formula is:

[0093]

[0094]

[0095] In the formula, Let x be the sum of squared errors of the j-th model, n represent the number of photovoltaic data points collected in a day, i represent the i-th point collected in a day, and x represent the sum of squared errors of the j-th model. i y represents the i-th actual value; i This represents the i-th predicted value; the method assigns high weights to models with small sum of squared errors.

[0096] On the other hand, refer to Figure 6 This invention provides a photovoltaic power generation prediction system 700, including a first module 710, a second module 720, a third module 730, a fourth module 740, a fifth module 750, and a sixth module 760. The first module 710 is used to perform weather factor correlation analysis based on historical load data of the photovoltaic power station and historical meteorological factor data to determine the target meteorological factor. The second module 720 is used to perform load sequence correlation analysis based on historical load data of the photovoltaic power station to determine the target specification of the load data. The third module 730 is used to construct a prediction neural network. The fourth module 740 is used to acquire historical load data of the target specification of the photovoltaic power station and historical data of the target meteorological factor associated with the historical load data of the target specification. The fifth module 750 is used to train the prediction neural network model based on the historical load data of the target specification and the historical data of the target meteorological factor to obtain a prediction model. The sixth module 760 is used to predict photovoltaic power generation based on the target meteorological factor data for the day to be predicted, using the prediction model.

[0097] The content of the method embodiments of the present invention is applicable to the system embodiments. The specific functions implemented in the system embodiments are the same as those in the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above methods.

[0098] Reference Figure 7 Another aspect of the present invention provides a photovoltaic power generation prediction device 800, including a processor 810 and a memory 820; the memory 810 is used to store a program; the processor 820 executes the program to implement the method described above.

[0099] The content of the method embodiments of the present invention is applicable to the device embodiments. The specific functions implemented by the device embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above methods.

[0100] Another aspect of this invention provides a computer-readable storage medium storing a program that is executed by a processor to implement the methods described above.

[0101] The content of the method embodiments of the present invention is applicable to the computer-readable storage medium embodiments. The specific functions implemented by the computer-readable storage medium embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above methods.

[0102] This invention also discloses a computer program product or computer program, which includes computer instructions stored in a computer-readable storage medium. A processor of a computer device can read the computer instructions from the computer-readable storage medium and execute the computer instructions, causing the computer device to perform the aforementioned method.

[0103] In summary, this invention uses historical data from photovoltaic power plants as input dimensions, taking weather factors affecting photovoltaic output as the input dimension, and trains the model using common artificial intelligence algorithms (linear regression, logistic regression, multinomial Naive Bayes, KNN nearest neighbors, SVM, GBDT iterative decision tree, random forest, and neural networks). First, feature analysis of weather factors affecting photovoltaic output is performed to determine the input feature vector. Second, historical photovoltaic data is preprocessed to segment a suitable data model, which is then fed into the artificial intelligence algorithm for model training. Third, photovoltaic power generation is predicted by incorporating relevant weather forecast factors. Finally, error analysis is performed on the prediction results, and negative feedback is used to train the model, improving the accuracy of photovoltaic power generation prediction.

[0104] In some alternative embodiments, the functions / operations mentioned in the block diagrams may not occur in the order shown in the operation diagrams. For example, depending on the functions / operations involved, two consecutively shown blocks may actually be executed substantially simultaneously, or the blocks may sometimes be executed in reverse order. Furthermore, the embodiments presented and described in the flowcharts of this invention are provided by way of example to provide a more comprehensive understanding of the technology. The disclosed methods are not limited to the operations and logic flows presented herein. Alternative embodiments are contemplated in which the order of various operations is altered and sub-operations described as part of a larger operation are executed independently.

[0105] Furthermore, although the invention has been described in the context of functional modules, it should be understood that, unless otherwise stated, one or more of the described functions and / or features may be integrated into a single physical device and / or software module, or one or more functions and / or features may be implemented in a separate physical device or software module. It is also understood that a detailed discussion of the actual implementation of each module is unnecessary for understanding the invention. Rather, given the properties, functions, and internal relationships of the various functional modules in the apparatus disclosed herein, the actual implementation of the module will be understood within the scope of conventional skill of an engineer. Therefore, those skilled in the art can implement the invention as set forth in the claims using ordinary techniques without excessive experimentation. It is also understood that the specific concepts disclosed are merely illustrative and not intended to limit the scope of the invention, which is determined by the full scope of the appended claims and their equivalents.

[0106] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, essentially, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0107] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-included system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0108] More specific examples of computer-readable media (a non-exhaustive list) include: electrical connections (electronic devices) having one or more wires, portable computer disk drives (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which the program can be printed, since the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.

[0109] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0110] In the description of this specification, references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0111] Although embodiments of the invention have been shown and described, those skilled in the art will understand that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the claims and their equivalents.

[0112] The above is a detailed description of the preferred embodiments of the present invention, but the present invention is not limited to the embodiments described. Those skilled in the art can make various equivalent modifications or substitutions without departing from the spirit of the present invention, and these equivalent modifications or substitutions are all included within the scope defined by the claims of this application.

Claims

1. A method for predicting photovoltaic power generation, characterized in that, include: Based on the historical load data of photovoltaic power stations, and combined with historical meteorological data, a correlation analysis of weather factors is conducted to determine the target meteorological factors. Based on the historical load data of the photovoltaic power station, load sequence correlation analysis is performed to determine the target specifications for the load data; Construct a predictive neural network; Acquire historical load data of the photovoltaic power station for the target specifications, and historical data of the target meteorological factors associated with the historical load data of the target specifications; Based on the historical load data of the target specification and the historical data of the target meteorological factors, the prediction neural network is trained to obtain the prediction model; Based on the target meteorological factors data for the date to be predicted, photovoltaic power generation is predicted using the prediction model. In this process, historical load data and historical data of target meteorological factors are collected and used as inputs to artificial intelligence algorithms. KNN, random forest, decision tree, Bayesian algorithm and MLP algorithm are called in turn to predict the prediction results of different algorithm models. Individual algorithm error analysis and combined prediction error analysis are performed respectively, and the one with the smallest error analysis result is selected as the prediction model. Among them, the combined prediction error analysis adopts the inverse variance method: the sum of squared prediction errors is an indicator reflecting the prediction accuracy. The larger the sum of squared prediction errors, the lower the prediction accuracy of the algorithm model, and thus the lower its importance in the combined prediction. The importance is reflected in its weight w in the combined prediction model. j The calculation formula is: In the formula, Let be the sum of squared errors of the j-th algorithm model. This indicates the amount of photovoltaic data collected in one day; Indicates the number of times a day a sample is collected. point; Indicates the first One actual value; Indicates the first The method assigns high weights to models with small sums of squared errors; The step of determining target meteorological factors by performing correlation analysis of weather factors based on historical load data of photovoltaic power plants and historical meteorological factor data includes: Based on the historical load data of photovoltaic power stations and combined with historical meteorological data, the correlation analysis of each meteorological factor is carried out sequentially using the correlation coefficient calculation formula to determine the target meteorological factors affecting the load of photovoltaic power stations. The formula for calculating the correlation coefficient is as follows: In the formula, Represents the correlation coefficient. Indicates the total amount of data. Indicates the first Historical load data, This represents the average value of historical load data. Indicates the first Historical data for each meteorological factor This represents the average value of historical data for meteorological factors.

2. The photovoltaic power generation prediction method according to claim 1, characterized in that, Also includes: The historical load data of the target specification is preprocessed.

3. The photovoltaic power generation prediction method according to claim 2, characterized in that, The data preprocessing of the historical load data for the target specification includes: Missing values ​​in the historical load data of the target specification are filled in using a single-point completion method. And / or, use a multi-point completion method to fill in the missing values ​​of the historical load data of the target specification.

4. The photovoltaic power generation prediction method according to claim 2, characterized in that, The data preprocessing of the historical load data for the target specification includes: Based on a preset threshold, determine the glitch data of the historical load data of the target specification; The burr data is replaced according to a preset processing method.

5. The photovoltaic power generation prediction method according to claim 1, characterized in that, The step of performing load sequence correlation analysis based on historical load data of photovoltaic power plants to determine the target specifications for load data includes: Based on the historical load data of photovoltaic power plants, nearest neighbor correlation analysis and periodic correlation analysis of load sequences are performed to determine the target specifications of load data; The periodic correlation analysis includes daily periodic correlation analysis, weekly periodic analysis, and annual periodic analysis.

6. The photovoltaic power generation prediction method according to claim 1, characterized in that, Also includes: Based on the load forecast and actual load values ​​from the prediction model, the model parameters of the prediction model are corrected using an error function.

7. A photovoltaic power generation prediction system, characterized in that, include: The first module is used to perform correlation analysis of weather factors based on historical load data of photovoltaic power plants and historical meteorological factor data to determine target meteorological factors. The second module is used to perform load sequence correlation analysis based on the historical load data of the photovoltaic power station to determine the target specifications of the load data; The third module is used to build the predictive neural network; The fourth module is used to acquire historical load data of the photovoltaic power station for the target specifications, and historical data of the target meteorological factors associated with the historical load data of the target specifications; The fifth module is used to train the prediction neural network based on the historical load data of the target specification and the historical data of the target meteorological factors to obtain a prediction model; The sixth module is used to predict photovoltaic power generation based on the target meteorological factors data for the date to be predicted, using the prediction model. In this process, historical load data and historical data of target meteorological factors are collected and used as inputs to artificial intelligence algorithms. KNN, random forest, decision tree, Bayesian algorithm and MLP algorithm are called in turn to predict the prediction results of different algorithm models. Individual algorithm error analysis and combined prediction error analysis are performed respectively, and the one with the smallest error analysis result is selected as the prediction model. Among them, the combined prediction error analysis adopts the inverse variance method: the sum of squared prediction errors is an indicator reflecting the prediction accuracy. The larger the sum of squared prediction errors, the lower the prediction accuracy of the algorithm model, and thus the lower its importance in the combined prediction. The importance is reflected in its weight w in the combined prediction model. j The calculation formula is: In the formula, Let be the sum of squared errors of the j-th algorithm model. This indicates the amount of photovoltaic data collected in one day; Indicates the number of times a day a sample is collected. point; Indicates the first One actual value; Indicates the first The method assigns high weights to models with small sums of squared errors; The step of determining target meteorological factors by performing correlation analysis of weather factors based on historical load data of photovoltaic power plants and historical meteorological factor data includes: Based on the historical load data of photovoltaic power stations and combined with historical meteorological data, the correlation analysis of each meteorological factor is carried out sequentially using the correlation coefficient calculation formula to determine the target meteorological factors affecting the load of photovoltaic power stations. The formula for calculating the correlation coefficient is as follows: In the formula, Represents the correlation coefficient. Indicates the total amount of data. Indicates the first Historical load data, This represents the average value of historical load data. Indicates the first Historical data for each meteorological factor This represents the average value of historical data for meteorological factors.

8. A photovoltaic power generation prediction device, characterized in that, Including the processor and memory; The memory is used to store programs; The processor executes the program to implement the method as described in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, The storage medium stores a program that is executed by a processor to implement the method as described in any one of claims 1 to 6.

Images