Distributed photovoltaic power generation intelligent prediction system and method

Abstract

The present application relates to the field of intelligent prediction technology, and specifically discloses a distributed photovoltaic power intelligent prediction system and method, which uses big data technology to perform correlation analysis on the historical power data and historical meteorological data of photovoltaic power generation equipment, digs out the correlation feature expression between photovoltaic power generation and meteorological data, and combines future meteorological data to predict future monthly power generation. In this way, it can provide a scientific basis for the operation and management of photovoltaic power stations, thereby improving power generation efficiency and economic benefits.

Description

Technical Field

[0001] The present application relates to the field of intelligent prediction technology, and more specifically, to a distributed photovoltaic power generation intelligent prediction system and method. Background Art

[0002] Distributed photovoltaic power generation refers to distributing photovoltaic power generation systems in multiple locations, converting solar energy into electricity and directly supplying it to local users or connecting it to the power grid. Compared with traditional centralized photovoltaic power generation, distributed photovoltaic power generation supplies electricity to users nearby, reducing transmission losses and grid construction costs. Users can directly use the electricity generated by the photovoltaic power generation system, reducing dependence on traditional power grids.

[0003] As the global demand for clean energy grows, photovoltaic power generation, as a renewable energy technology, has been increasingly widely used. However, the power generation of photovoltaic power stations is mainly affected by weather factors and the working status of photovoltaic components, such as ground irradiance, temperature, and performance of photovoltaic arrays at the location of the photovoltaic power station, and the relationship between power generation and weather factors is nonlinear. Therefore, how to accurately predict photovoltaic power generation has become an important challenge for power grid dispatchers.

[0004] Therefore, a distributed photovoltaic power generation intelligent prediction system and method are expected. Summary of the invention

[0005] In order to solve the above technical problems, this application is proposed. The embodiment of this application provides a distributed photovoltaic power generation intelligent prediction system and method, which uses big data technology to perform correlation analysis on the historical power generation data and historical meteorological data of photovoltaic power generation equipment, digs out the correlation feature expression between photovoltaic power generation and meteorological data, and combines future meteorological data to predict future monthly power generation. In this way, it can provide a scientific basis for the operation and management of photovoltaic power stations, thereby improving power generation efficiency and economic benefits.

[0006] Accordingly, according to one aspect of the present application, a distributed photovoltaic power generation intelligent prediction system is provided, which includes:

[0007] A data acquisition module is used to obtain meteorological data of each day in history, meteorological data of each day in the next month, and photovoltaic power generation values ​​of each day in history, wherein the meteorological data includes ground irradiance, humidity, temperature and wind speed;

[0008] A meteorological data time series correlation coding module is used to extract meteorological time series correlation features from the meteorological data of each historical day and the meteorological data of each day in the next month to obtain a historical meteorological time series correlation feature vector and a future meteorological time series correlation feature vector;

[0009] A power time series encoding module is used to extract time series features of the photovoltaic power generation values ​​of each historical day to obtain a historical power generation feature vector;

[0010] A historical data association coding module, used for performing association coding on the historical meteorological time series association feature vector and the historical power generation feature vector to obtain a historical meteorological-power generation association feature matrix;

[0011] The photovoltaic power generation prediction module is used to determine the photovoltaic power generation power prediction value in the future month based on the interaction characteristics between the future meteorological time series correlation feature vector and the historical meteorological-power generation power correlation feature matrix.

[0012] In the above-mentioned distributed photovoltaic power generation intelligent prediction system, the meteorological data time series association coding module includes: a matrixing unit, which is used to arrange the meteorological data of each historical day and the meteorological data of each day in the next month into a historical meteorological matrix and a future meteorological matrix according to the time dimension and the meteorological sample dimension respectively; a meteorological data time series association coding unit, which is used to pass the historical meteorological matrix and the future meteorological matrix through a meteorological time series association feature extractor based on a convolutional neural network model to obtain the historical meteorological time series association feature vector and the future meteorological time series association feature vector.

[0013] In the above-mentioned distributed photovoltaic power generation intelligent prediction system, the power time series encoding module is used to: arrange the photovoltaic power generation power values ​​of each historical day into a power input vector according to the time dimension and then obtain the historical power generation power feature vector through a power time series encoder based on a one-dimensional convolutional layer.

[0014] In the above-mentioned distributed photovoltaic power generation intelligent prediction system, the historical data association coding module is used to: associate and encode the historical meteorological time series association feature vector and the historical power generation feature vector based on the Gaussian density map to obtain the historical meteorological-power generation association feature matrix.

[0015] In the above-mentioned distributed photovoltaic power generation intelligent prediction system, the historical data association coding module includes: a Gaussian density map construction unit, which is used to use the Gaussian density map to perform feature association on the historical meteorological time series association feature vector and the historical power generation feature vector using the following association formula to obtain an associated Gaussian density map; wherein the association formula is: Wherein, μ represents the position mean vector between the historical meteorological time series associated feature vector and the historical power generation feature vector, and the value of each position of σ represents the variance between the feature values ​​of each position in the historical meteorological time series associated feature vector and the historical power generation feature vector, represents a Gaussian density probability function, x represents a variable of a Gaussian density map; and a Gaussian discretization unit, which is used to discretize the Gaussian distribution of each position of the associated Gaussian density map to obtain the historical meteorological-power generation correlation feature matrix.

[0016] In the above-mentioned distributed photovoltaic power intelligent prediction system, the photovoltaic power prediction module includes: a query unit, which is used to multiply the historical meteorological-power generation correlation feature matrix and the future meteorological time series correlation feature vector to obtain a monthly photovoltaic power prediction feature vector; an optimization unit, which is used to perform manifold hyperconvex correlation derivative representation optimization on the monthly photovoltaic power prediction feature vector to obtain an optimized monthly photovoltaic power prediction feature vector; a prediction result generation unit, which is used to pass the optimized monthly photovoltaic power prediction feature vector through a decoder to obtain a decoded value, and the decoded value is used to represent the photovoltaic power prediction value of the future month.

[0017] In the above-mentioned distributed photovoltaic power generation intelligent prediction system, the optimization unit is used to: perform manifold hyperconvex correlation derivative representation optimization on the monthly photovoltaic power generation prediction feature vector according to the following optimization formula to obtain the optimized monthly photovoltaic power generation prediction feature vector; wherein the optimization formula is:

[0018]

[0019] in, represents the monthly photovoltaic power generation prediction feature vector, v i represents the eigenvalue of the ith position of the monthly photovoltaic power prediction eigenvector, v j represents the eigenvalue of the jth position of the monthly photovoltaic power prediction feature vector, α and β represent weight hyperparameters, and v′ i Represents the eigenvalue of the i-th position of the optimized monthly photovoltaic power generation prediction eigenvector.

[0020] In the above-mentioned distributed photovoltaic power generation intelligent prediction system, the prediction result generation unit is used to: use the decoder to decode and regress the optimized monthly photovoltaic power generation prediction feature vector using the following decoding formula to obtain the decoded value; wherein the decoding formula is: Where X is the optimized monthly photovoltaic power generation power prediction feature vector, Y is the decoded value, W is the weight matrix, Represents matrix multiplication.

[0021] According to another aspect of the present application, a distributed photovoltaic power generation intelligent prediction method is provided, which includes:

[0022] Obtain meteorological data for each day in history, meteorological data for each day in the next month, and photovoltaic power generation values ​​for each day in history, wherein the meteorological data includes ground irradiance, humidity, temperature, and wind speed;

[0023] Extract meteorological time series correlation features from the meteorological data of each historical day and the meteorological data of each day in the next month respectively to obtain a historical meteorological time series correlation feature vector and a future meteorological time series correlation feature vector;

[0024] Extracting time series features of photovoltaic power generation values ​​on each of the historical days to obtain a historical power generation feature vector;

[0025] Performing association coding on the historical meteorological time series correlation feature vector and the historical power generation feature vector to obtain a historical meteorological-power generation correlation feature matrix;

[0026] Based on the interaction characteristics between the future meteorological time series correlation feature vector and the historical meteorological-power generation power correlation feature matrix, the photovoltaic power generation power forecast value for the future month is determined.

[0027] Compared with the prior art, the distributed photovoltaic power generation intelligent prediction system and method provided by this application uses big data technology to perform correlation analysis on the historical power generation data and historical meteorological data of photovoltaic power generation equipment, digs out the correlation feature expression between photovoltaic power generation and meteorological data, and combines future meteorological data to predict future monthly power generation. In this way, it can provide a scientific basis for the operation and management of photovoltaic power stations, thereby improving power generation efficiency and economic benefits. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0029] Figure 1 It is a block diagram of a distributed photovoltaic power generation intelligent prediction system according to an embodiment of the present application.

[0030] Figure 2 Schematic diagram of the architecture of a distributed photovoltaic power generation intelligent prediction system according to an embodiment of the present application.

[0031] Figure 3 It is a block diagram of a meteorological data time series association coding module in a distributed photovoltaic power generation intelligent prediction system according to an embodiment of the present application.

[0032] Figure 4 It is a block diagram of a historical data association coding module in a distributed photovoltaic power generation intelligent prediction system according to an embodiment of the present application.

[0033] Figure 5 It is a block diagram of a photovoltaic power generation prediction module in a distributed photovoltaic power generation intelligent prediction system according to an embodiment of the present application.

[0034] Figure 6 The present invention is a flowchart of a distributed photovoltaic power generation intelligent prediction method according to an embodiment of the present application. DETAILED DESCRIPTION

[0035] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described here.

[0036] Figure 1 It is a block diagram of a distributed photovoltaic power generation intelligent prediction system according to an embodiment of the present application.

[0037] Figure 2 FIG. 1 is a schematic diagram of the architecture of a distributed photovoltaic power generation intelligent prediction system according to an embodiment of the present application. Figure 1 and Figure 2 As shown, according to the embodiment of the present application, the distributed photovoltaic power generation intelligent prediction system 100 includes: a data acquisition module 110, which is used to obtain meteorological data of each historical day, meteorological data of each day in the next month, and photovoltaic power generation power values ​​of each historical day, wherein the meteorological data includes ground irradiance, humidity, temperature and wind speed; a meteorological data time series association coding module 120, which is used to extract meteorological time series association features of the meteorological data of each historical day and the meteorological data of each day in the next month respectively to obtain a historical meteorological time series association feature vector and a future meteorological time series association feature vector; a power time series coding module 130, which is used to extract time series features of the photovoltaic power generation power values ​​of each historical day to obtain a historical power generation feature vector; a historical data association coding module 140, which is used to associate and encode the historical meteorological time series association feature vector and the historical power generation feature vector to obtain a historical meteorological-power generation power association feature matrix; a photovoltaic power generation power prediction module 150, which is used to determine the photovoltaic power generation power prediction value for the future month based on the interaction features between the future meteorological time series association feature vector and the historical meteorological-power generation power association feature matrix.

[0038] As mentioned in the background technology above, weather factors are one of the important factors affecting photovoltaic power generation. Photovoltaic power stations rely on sunlight to generate electricity, so weather conditions have a direct impact on power generation. On sunny days, there is plenty of sunshine, and photovoltaic power stations can exert their maximum power generation capacity; on cloudy or overcast days, the light intensity decreases, and the power generation will also decrease accordingly. However, traditional photovoltaic power generation prediction methods usually use linear models for prediction. Since photovoltaic power generation is affected by a combination of multiple factors, including light intensity, temperature, humidity, etc., the relationship between these weather factors is often nonlinear. Therefore, this linear model-based prediction method often fails to capture the complex nonlinear relationship and dynamic changes between power generation and weather factors, and it is difficult to achieve accurate prediction of photovoltaic power generation.

[0039] In response to the above technical problems, the technical concept of this application is to use big data technology to perform correlation analysis on the historical power generation data and historical meteorological data of photovoltaic power generation equipment, to mine the correlation feature expression between photovoltaic power generation and meteorological data, and to combine future meteorological data to predict future monthly power generation. In this way, it can provide a scientific basis for the operation and management of photovoltaic power stations, thereby improving power generation efficiency and economic benefits.

[0040] In the above-mentioned distributed photovoltaic power intelligent prediction system 100, the data acquisition module 110 is used to obtain the meteorological data of each historical day, the meteorological data of each day in the next month, and the photovoltaic power generation value of each historical day, wherein the meteorological data includes ground irradiance, humidity, temperature and wind speed. It should be understood that there is a certain correlation between photovoltaic power generation and meteorological data. Factors such as ground irradiance, humidity, temperature and wind speed in meteorological data will directly affect the power generation efficiency of the photovoltaic power generation system. Irradiance is the main driving factor of photovoltaic power generation, humidity and temperature will affect the temperature and conductivity of photovoltaic cells, and wind speed will affect the heat dissipation effect of photovoltaic modules. Therefore, by obtaining the meteorological data of each historical day and the photovoltaic power generation value of each historical day, the two can be further analyzed for correlation to reveal the correlation between photovoltaic power generation and meteorological factors. That is to say, in the technical solution of the present application, the meteorological data of each historical day and the photovoltaic power generation value are used to construct a correlation model between photovoltaic power generation and meteorological data, and capture the impact of meteorological data on photovoltaic power generation. The meteorological data of each day in the next month is used to predict the future photovoltaic power generation according to the correlation between photovoltaic power generation and meteorological data.

[0041] In the above-mentioned distributed photovoltaic power generation intelligent prediction system 100, the meteorological data time series correlation coding module 120 is used to extract meteorological time series correlation features from the meteorological data of each historical day and the meteorological data of each day in the next month to obtain a historical meteorological time series correlation feature vector and a future meteorological time series correlation feature vector. Specifically, Figure 3 FIG. 1 is a block diagram of a meteorological data time series correlation coding module in a distributed photovoltaic power generation intelligent prediction system according to an embodiment of the present application. Figure 3 As shown, the meteorological data time series association coding module 120 includes: a matrixing unit 121, which is used to arrange the meteorological data of each historical day and the meteorological data of each day in the next month into a historical meteorological matrix and a future meteorological matrix according to the time dimension and the meteorological sample dimension respectively; a meteorological data time series association coding unit 122, which is used to pass the historical meteorological matrix and the future meteorological matrix through a meteorological time series association feature extractor based on a convolutional neural network model to obtain the historical meteorological time series association feature vector and the future meteorological time series association feature vector.

[0042] Specifically, the matrixing unit 121 is used to arrange the meteorological data of each day in the history and the meteorological data of each day in the future month into a historical meteorological matrix and a future meteorological matrix according to the time dimension and the meteorological sample dimension, respectively. It should be understood that by arranging the meteorological data according to the time dimension, the time sequence information of the meteorological data can be retained, which is convenient for subsequent capture of the changing law of the meteorological data of each day in the time dimension. By arranging the meteorological data according to the meteorological sample dimension, the data of different meteorological factors (such as ground irradiance, humidity, temperature and wind speed) at the same time point can be organized together to form a meteorological sample, and the correlation between different meteorological factors is taken into consideration, so that the influence of multiple meteorological factors on photovoltaic power generation can be considered simultaneously in the subsequent feature association analysis process. In other words, by arranging the meteorological data of each day into a meteorological matrix according to the time dimension and the meteorological sample dimension, the changing characteristics of the meteorological data in the time dimension and the correlation characteristics between different meteorological factors can be captured in the subsequent data processing process, providing useful data input for subsequent feature extraction and association analysis.

[0043] Specifically, the meteorological data time series association encoding unit 122 is used to pass the historical meteorological matrix and the future meteorological matrix through a meteorological time series association feature extractor based on a convolutional neural network model to obtain the historical meteorological time series association feature vector and the future meteorological time series association feature vector. It should be understood that in order to capture the temporal variation pattern of meteorological data, a meteorological time series association feature extractor based on a convolutional neural network model is further used to extract time series association features from the historical meteorological matrix and the future meteorological matrix. Those of ordinary skill in the art should know that convolutional neural networks have good feature extraction capabilities when processing two-dimensional data. The convolution layer of a convolutional neural network can capture local correlation relationships in the input data by performing convolution operations on the input data using a set of learnable convolution kernels. In the technical solution of the present application, the historical meteorological matrix and the future meteorological matrix are processed by a meteorological time series correlation feature extractor based on a convolutional neural network model. The meteorological time series correlation feature extractor can use a convolution layer to perform sliding convolution operations of different time windows on the historical meteorological matrix and the future meteorological matrix, respectively, to capture the dynamic change law of meteorological data in the time dimension. The features output by the convolution layer are downsampled through a pooling layer to reduce the size of the features, while reducing the amount of calculation and the number of parameters, retaining the important feature representation of the meteorological data, and providing valuable data input for subsequent correlation coding and photovoltaic power generation prediction tasks.

[0044] In the above-mentioned distributed photovoltaic power generation intelligent prediction system 100, the power time series encoding module 130 is used to extract the time series features of the photovoltaic power generation values ​​of each historical day to obtain the historical power generation feature vector. In a specific example of the present application, the encoding method of extracting the time series features of the photovoltaic power generation values ​​of each historical day to obtain the historical power generation feature vector is to arrange the photovoltaic power generation values ​​of each historical day into a power input vector according to the time dimension and then pass it through a power time series encoder based on a one-dimensional convolution layer to obtain the historical power generation feature vector. It should be understood that photovoltaic power generation is a sequence data that changes with time. By arranging the photovoltaic power generation values ​​of each historical day into a power input vector according to the time dimension, the dynamic change information of the photovoltaic power generation value in the time dimension can be retained. Then, the power input vector is processed using a power time series encoder based on a one-dimensional convolution layer. The one-dimensional convolution layer can automatically learn the local correlation features in different time windows on the power input vector by applying a one-dimensional convolution kernel to perform a sliding convolution operation on the power input vector, and capture the temporal change law of the power data. Through the combination of multiple convolution kernels, abstract features at different levels are extracted, including short-term changes and long-term trends of photovoltaic power generation, so as to capture the dynamic change pattern of photovoltaic power generation and provide input for subsequent correlation analysis and prediction.

[0045] In the above-mentioned distributed photovoltaic power generation intelligent prediction system 100, the historical data association coding module 140 is used to associate the historical meteorological time series association feature vector and the historical power generation feature vector to obtain a historical meteorological-power generation association feature matrix. It should be understood that there is a certain correlation between photovoltaic power generation and meteorological data. In order to capture the dynamic correlation between photovoltaic power generation and meteorological data, the historical meteorological time series association feature vector and the historical power generation feature vector are further associated. Encoding. In a specific example of the present application, the encoding method of associating the historical meteorological time series association feature vector and the historical power generation feature vector to obtain the historical meteorological-power generation association feature matrix is ​​based on a Gaussian density map to associate the historical meteorological time series association feature vector and the historical power generation feature vector to obtain the historical meteorological-power generation association feature matrix. Ordinary technicians in this field should know that Gaussian density map is a commonly used probability distribution map that can be used to model the association relationship of data. In the technical solution of the present application, the historical meteorological time series correlation feature vector and the historical power generation feature vector are associated and encoded based on the Gaussian density map, and the historical meteorological time series correlation feature vector and the historical power generation feature vector can be mapped to a shared feature space, so that the historical meteorological time series correlation feature vector and the historical power generation feature vector are aligned in the feature space, and the correlation between the historical meteorological time series correlation feature vector and the historical power generation feature vector is calculated based on the probability distribution of the historical meteorological time series correlation feature vector and the historical power generation feature vector in the feature space, so as to capture the power generation change law under different meteorological conditions, reflect the long-term impact of meteorological factors on photovoltaic power generation, and provide a reference for future power generation prediction.

[0046] Figure 4 FIG. 1 is a block diagram of a historical data association coding module in a distributed photovoltaic power generation intelligent prediction system according to an embodiment of the present application. Figure 4 As shown, the historical data association coding module 140 includes: a Gaussian density map construction unit 141, which is used to use the Gaussian density map to perform feature association on the historical meteorological time series association feature vector and the historical power generation feature vector using the following association formula to obtain an associated Gaussian density map; wherein the association formula is: Wherein, μ represents the position mean vector between the historical meteorological time series associated feature vector and the historical power generation feature vector, and the value of each position of σ represents the variance between the feature values ​​of each position in the historical meteorological time series associated feature vector and the historical power generation feature vector, represents a Gaussian density probability function, x represents a variable of the Gaussian density map; and a Gaussian discretization unit 142, which is used to discretize the Gaussian distribution of each position of the associated Gaussian density map to obtain the historical meteorological-power generation power correlation feature matrix.

[0047] In the above-mentioned distributed photovoltaic power generation intelligent prediction system 100, the photovoltaic power generation prediction module 150 is used to determine the photovoltaic power generation power prediction value for the future month based on the interaction characteristics between the future meteorological time series correlation feature vector and the historical meteorological-power generation correlation feature matrix. Specifically, Figure 5 FIG. 1 is a block diagram of a photovoltaic power generation prediction module in a distributed photovoltaic power generation intelligent prediction system according to an embodiment of the present application. Figure 5 As shown, the photovoltaic power prediction module 150 includes: a query unit 151, which is used to multiply the historical meteorological-power generation correlation feature matrix and the future meteorological time series correlation feature vector to obtain a monthly photovoltaic power generation prediction feature vector; an optimization unit 152, which is used to perform manifold hyperconvex correlation derivative representation optimization on the monthly photovoltaic power generation prediction feature vector to obtain an optimized monthly photovoltaic power generation prediction feature vector; a prediction result generation unit 153, which is used to pass the optimized monthly photovoltaic power generation prediction feature vector through a decoder to obtain a decoded value, and the decoded value is used to represent the photovoltaic power prediction value of the future month.

[0048] Specifically, the query unit 151 is used to multiply the historical meteorological-power generation correlation feature matrix and the future meteorological time series correlation feature vector to obtain the monthly photovoltaic power generation prediction feature vector. It should be understood that the historical meteorological-power generation correlation feature matrix reflects the correlation between historical meteorological data and power generation data, reveals the degree of influence of different meteorological conditions on photovoltaic power generation, and establishes a relationship model between meteorological conditions and power generation. The future meteorological time series correlation feature vector contains the characteristics of meteorological data in the future period of time. In order to combine the correlation characteristics of historical meteorological data and power generation with future meteorological data to generate a prediction of future photovoltaic power generation, the future meteorological time series correlation feature vector is further multiplied with the historical meteorological-power generation correlation feature matrix. That is, through the multiplication operation, the weight information of the historical meteorological-power generation correlation feature matrix is ​​used to adjust the influence of future meteorological characteristics, so as to capture the influence of future meteorological time series correlation characteristics on future power generation, and apply it to the prediction process to provide more accurate data input for the prediction of photovoltaic power generation.

[0049] Specifically, the optimization unit 152 is used to perform manifold hyperconvex correlation derivative representation optimization on the monthly photovoltaic power prediction feature vector to obtain an optimized monthly photovoltaic power prediction feature vector. In particular, in the technical solution of the present application, the monthly photovoltaic power prediction feature vector is obtained by multiplying the historical meteorological-power generation correlation feature matrix and the future meteorological feature vector. However, due to the data manifold complexity of the high-dimensional feature space, the eigenvalues ​​at each position in the monthly photovoltaic power prediction feature vector have a technical problem of distribution discretization for the confidence of the target class probability label domain. Similar to the aforementioned problem, the discretization of data distribution in the high-dimensional feature space is one of the main reasons for the poor information aggregation. In the high-dimensional space, the distance and angle relationship between data points is more complex, and the distribution of data points is more sparse and discretized. This leads to the problem of discretization of the distribution of the eigenvalues ​​at each position in the monthly photovoltaic power prediction feature vector in the target class probability label domain, that is, the correlation between the eigenvalues ​​is weak, and it is difficult to accurately represent the confidence of the target class. In addition, there may be information loss or redundancy in the feature fusion process. When multiplying the historical meteorological-power generation correlation feature matrix and the future meteorological feature vector, the correlation information between the two may not be fully retained, resulting in a decrease in the information aggregation degree in the feature vector. Based on this, considering that the monthly photovoltaic power generation prediction feature vector has a problem of poor information aggregation in the data manifold of the high-dimensional feature space, that is, the eigenvalues ​​at each position in the monthly photovoltaic power generation prediction feature vector have a technical problem of distribution discretization for the confidence of the target class probability label domain, in the technical solution of the present application, the monthly photovoltaic power generation prediction feature vector is optimized by manifold hyperconvex correlation derivative representation.

[0050] In a specific example of the present application, the optimization unit 152 is used to: perform manifold hyperconvex correlation derivative representation optimization on the monthly photovoltaic power generation prediction feature vector using the following optimization formula to obtain the optimized monthly photovoltaic power generation prediction feature vector; wherein the optimization formula is:

[0051]

[0052] in, represents the monthly photovoltaic power generation prediction feature vector, v i represents the eigenvalue of the ith position of the monthly photovoltaic power prediction eigenvector, v j represents the eigenvalue of the jth position of the monthly photovoltaic power prediction feature vector, α and β represent weight hyperparameters, and v′ i Represents the eigenvalue of the i-th position of the optimized monthly photovoltaic power generation prediction eigenvector.

[0053] That is, considering that the monthly photovoltaic power generation prediction feature vector has a poor information aggregation problem in the data manifold of the high-dimensional feature space, that is, the confidence of the feature values ​​at each position in the monthly photovoltaic power generation prediction feature vector for the target class probability label domain has a technical problem of distribution discretization, in the technical solution of the present application, the monthly photovoltaic power generation prediction feature vector is optimized by manifold hyperconvex correlation derivative representation, which constructs a normalization function based on the robustness of information minimization loss, so that the feature values ​​at each position in the feature vector can maintain hyperconvex correlation in its sub-dimension, thereby realizing information clustering of the feature vector. In this way, the information redundancy and noise in the feature vector are effectively suppressed, the information aggregation of the feature vector is enhanced, and the confidence of the feature vector is improved.

[0054] Specifically, the prediction result generating unit 153 is used to pass the optimized monthly photovoltaic power prediction feature vector through a decoder to obtain a decoded value, and the decoded value is used to represent the photovoltaic power prediction value of the future month. It should be understood that the decoder is a neural network model, which can perform an inverse transformation on the input feature representation through a series of reverse calculation operations and map it back to the original data space. In the technical solution of the present application, the decoder is used to map the optimized monthly photovoltaic power prediction feature vector back to the original prediction target space, that is, the numerical range of photovoltaic power. In other words, the optimized monthly photovoltaic power prediction feature vector is converted into a specific prediction value by the decoder to obtain a specific prediction of the future monthly photovoltaic power, so that the prediction result has an actual power numerical meaning. In this way, the operation and management of the photovoltaic power generation system can be guided based on the photovoltaic power prediction value of the future month, such as reasonably arranging power generation plans, optimizing power grid scheduling, etc.

[0055] In a specific example of the present application, the prediction result generating unit 153 is used to: use the decoder to decode and regress the optimized monthly photovoltaic power generation power prediction feature vector using the following decoding formula to obtain the decoded value; wherein the decoding formula is: Where X is the optimized monthly photovoltaic power generation power prediction feature vector, Y is the decoded value, W is the weight matrix, Represents matrix multiplication.

[0056] In summary, the distributed photovoltaic power generation intelligent prediction system according to the embodiment of the present application is explained, which uses big data technology to perform correlation analysis on the historical power generation data and historical meteorological data of photovoltaic power generation equipment, digs out the correlation feature expression between photovoltaic power generation and meteorological data, and combines future meteorological data to perform future monthly power generation prediction. In this way, it can provide a scientific basis for the operation and management of photovoltaic power stations, thereby improving power generation efficiency and economic benefits.

[0057] Figure 6 FIG. 1 is a flow chart of a distributed photovoltaic power generation intelligent prediction method according to an embodiment of the present application. Figure 6 As shown, according to the embodiment of the present application, the distributed photovoltaic power generation intelligent prediction method includes the following steps: S110, obtaining meteorological data of each historical day, meteorological data of each day in the next month and photovoltaic power generation values ​​of each historical day, wherein the meteorological data include ground irradiance, humidity, temperature and wind speed; S120, respectively performing meteorological time series correlation feature extraction on the meteorological data of each historical day and the meteorological data of each day in the next month to obtain a historical meteorological time series correlation feature vector and a future meteorological time series correlation feature vector; S130, performing time series feature extraction on the photovoltaic power generation values ​​of each historical day to obtain a historical power generation feature vector; S140, performing association coding on the historical meteorological time series correlation feature vector and the historical power generation feature vector to obtain a historical meteorological-power generation correlation feature matrix; S150, determining the photovoltaic power generation power forecast value for the future month based on the interaction features between the future meteorological time series correlation feature vector and the historical meteorological-power generation correlation feature matrix.

[0058] Here, those skilled in the art can understand that the specific operations of each step in the above-mentioned distributed photovoltaic power generation intelligent prediction method have been referred to above. Figures 1 to 5 The description of the distributed photovoltaic power generation intelligent prediction system has been introduced in detail, and therefore, its repeated description will be omitted.

[0059] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described or recorded in detail in a certain embodiment, please refer to the relevant description of other embodiments. In the several embodiments provided by the present invention, it should be understood that the disclosed devices, systems and methods can be implemented in other ways. For example, the system embodiments described above are only schematic. For example, the module division is only a logical function division, and there may be other division methods in actual implementation. The modules described as separate components may or may not be physically separated, and the components displayed as modules may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment.

[0060] It will be apparent to those skilled in the art that the invention is not limited to the details of the exemplary embodiments described above and that the invention can be implemented in other specific forms without departing from the spirit or essential features of the invention. Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description, and it is intended that all variations falling within the meaning and scope of the equivalent elements of the claims be included in the invention. Any reference to a figure in a claim should not be considered as limiting the claim to which it relates.

[0061] In addition, it is obvious that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. The multiple units stated in the system claims can also be implemented by one unit through software or hardware.

[0062] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the spirit and scope of the technical solution of the present invention.

Claims

1. A distributed photovoltaic power generation intelligent prediction system, characterized in that: include: A data acquisition module is used to obtain meteorological data of each day in history, meteorological data of each day in the next month, and photovoltaic power generation values ​​of each day in history, wherein the meteorological data includes ground irradiance, humidity, temperature and wind speed; A meteorological data time series correlation coding module is used to extract meteorological time series correlation features from the meteorological data of each historical day and the meteorological data of each day in the next month to obtain a historical meteorological time series correlation feature vector and a future meteorological time series correlation feature vector; A power time series encoding module is used to extract time series features of the photovoltaic power generation values ​​of each historical day to obtain a historical power generation feature vector; A historical data association coding module, used for performing association coding on the historical meteorological time series association feature vector and the historical power generation feature vector to obtain a historical meteorological-power generation association feature matrix; A photovoltaic power generation prediction module is used to determine the photovoltaic power generation prediction value for the future month based on the interaction characteristics between the future meteorological time series correlation feature vector and the historical meteorological-power generation correlation feature matrix; Wherein, the photovoltaic power generation power prediction module includes: A query unit, used for multiplying the historical meteorological-power generation correlation feature matrix and the future meteorological time series correlation feature vector to obtain a monthly photovoltaic power generation power prediction feature vector; An optimization unit, used for performing manifold hyperconvex correlation derivative representation optimization on the monthly photovoltaic power generation prediction feature vector to obtain an optimized monthly photovoltaic power generation prediction feature vector; A prediction result generating unit, used for passing the optimized monthly photovoltaic power generation prediction feature vector through a decoder to obtain a decoded value, wherein the decoded value is used to represent the photovoltaic power generation prediction value of the future month; The optimization unit is used to perform manifold hyperconvex correlation derivative representation optimization on the monthly photovoltaic power prediction feature vector using the following optimization formula to obtain the optimized monthly photovoltaic power prediction feature vector; wherein the optimization formula is: in, represents the monthly photovoltaic power generation power prediction feature vector, The first character vector representing the monthly photovoltaic power generation prediction The eigenvalues ​​at the positions, The first character vector representing the monthly photovoltaic power generation prediction The eigenvalues ​​at the positions, and represents the weight hyperparameter, The first The feature value at each position.

2. The distributed photovoltaic power generation intelligent prediction system according to claim 1 is characterized in that: The meteorological data temporal association coding module includes: A matrixing unit, used for arranging the meteorological data of each historical day and the meteorological data of each day in the next month into a historical meteorological matrix and a future meteorological matrix according to a time dimension and a meteorological sample dimension respectively; The meteorological data time series correlation encoding unit is used to pass the historical meteorological matrix and the future meteorological matrix through a meteorological time series correlation feature extractor based on a convolutional neural network model to obtain the historical meteorological time series correlation feature vector and the future meteorological time series correlation feature vector.

3. The distributed photovoltaic power generation intelligent prediction system according to claim 2 is characterized in that: The power timing encoding module is used to: The photovoltaic power generation power values ​​of the historical days are arranged as power input vectors according to the time dimension and then passed through a power time series encoder based on a one-dimensional convolutional layer to obtain the historical power generation feature vector.

4. The distributed photovoltaic power generation intelligent prediction system according to claim 3 is characterized in that: The historical data association coding module is used to: The historical meteorological time series correlation feature vector and the historical power generation feature vector are associated and encoded based on the Gaussian density map to obtain the historical meteorological-power generation correlation feature matrix.

5. The distributed photovoltaic power generation intelligent prediction system according to claim 4 is characterized in that: The historical data association coding module includes: A Gaussian density map construction unit, used to use the Gaussian density map to perform feature association on the historical meteorological time series associated feature vector and the historical power generation feature vector using the following association formula to obtain an associated Gaussian density map; Wherein, the association formula is: in, represents the positional mean vector between the historical meteorological time series correlation feature vector and the historical power generation feature vector, and The value of each position of represents the variance between the characteristic values ​​of each position in the historical meteorological time series correlation characteristic vector and the historical power generation characteristic vector, represents the Gaussian density probability function, Variable representing the Gaussian density map; The Gaussian discretization unit is used to discretize the Gaussian distribution of each position of the associated Gaussian density map to obtain the historical meteorological-power generation power correlation characteristic matrix.

6. The distributed photovoltaic power generation intelligent prediction system according to claim 5 is characterized in that: The prediction result generating unit is used to: use the decoder to perform decoding regression on the optimized monthly photovoltaic power generation power prediction feature vector using the following decoding formula to obtain the decoding value; wherein the decoding formula is: ,in is the optimized monthly photovoltaic power prediction feature vector, is the decoded value, is the weight matrix, Represents matrix multiplication.

7. A distributed photovoltaic power generation intelligent prediction method, characterized in that: include: Obtain meteorological data for each day in history, meteorological data for each day in the next month, and photovoltaic power generation values ​​for each day in history, wherein the meteorological data includes ground irradiance, humidity, temperature, and wind speed; Extract meteorological time series correlation features from the meteorological data of each historical day and the meteorological data of each day in the next month respectively to obtain a historical meteorological time series correlation feature vector and a future meteorological time series correlation feature vector; Extracting time series features of photovoltaic power generation values ​​on each of the historical days to obtain a historical power generation feature vector; Performing association coding on the historical meteorological time series correlation feature vector and the historical power generation feature vector to obtain a historical meteorological-power generation correlation feature matrix; Determine the photovoltaic power generation power forecast value for the future month based on the interaction characteristics between the future meteorological time series correlation feature vector and the historical meteorological-power generation power correlation feature matrix; Wherein, based on the interaction characteristics between the future meteorological time series correlation feature vector and the historical meteorological-power generation power correlation feature matrix, the photovoltaic power generation power forecast value for the future month is determined, including: Multiplying the historical meteorological-power generation correlation feature matrix and the future meteorological time series correlation feature vector to obtain a monthly photovoltaic power generation power prediction feature vector; Performing manifold hyperconvex correlation derivative representation optimization on the monthly photovoltaic power generation prediction feature vector to obtain an optimized monthly photovoltaic power generation prediction feature vector; Passing the optimized monthly photovoltaic power generation prediction feature vector through a decoder to obtain a decoded value, wherein the decoded value is used to represent the photovoltaic power generation prediction value for the future month; The monthly photovoltaic power generation prediction feature vector is optimized by manifold hyperconvex correlation derivative representation to obtain an optimized monthly photovoltaic power generation prediction feature vector, including: the monthly photovoltaic power generation prediction feature vector is optimized by manifold hyperconvex correlation derivative representation to obtain the optimized monthly photovoltaic power generation prediction feature vector according to the following optimization formula; wherein the optimization formula is: in, represents the monthly photovoltaic power generation power prediction feature vector, The first character vector representing the monthly photovoltaic power generation prediction The eigenvalues ​​at the positions, The first character vector representing the monthly photovoltaic power generation prediction The eigenvalues ​​at the positions, and represents the weight hyperparameter, The first The feature value at each position.

8. The distributed photovoltaic power generation intelligent prediction method according to claim 7 is characterized in that: Extracting meteorological time series correlation features from the meteorological data of each historical day and the meteorological data of each day in the next month respectively to obtain a historical meteorological time series correlation feature vector and a future meteorological time series correlation feature vector, including: Arrange the meteorological data of each historical day and the meteorological data of each day in the next month into a historical meteorological matrix and a future meteorological matrix according to the time dimension and the meteorological sample dimension respectively; The historical meteorological matrix and the future meteorological matrix are respectively passed through a meteorological time series correlation feature extractor based on a convolutional neural network model to obtain the historical meteorological time series correlation feature vector and the future meteorological time series correlation feature vector.

Images