Wind power probability prediction method

Abstract

The invention provides a wind power probability prediction method, which comprises the following steps: calculating a Moran index value of an area where a wind power plant cluster is located, extracting spatial features, and forming a final input matrix; decomposing the wind power time sequence data to obtain an initial component, and decomposing the initial component to obtain a plurality of components; calculating an entropy value by adopting a fine composite multi-scale weighted permutation entropy, and performing comprehensive measurement and analysis by adopting an averaging method to obtain a plurality of comprehensive components; according to the FAAM-Stacking model, a final deterministic prediction result is obtained, and the performance is verified; and according to the error of the final deterministic prediction result, a distributed kernel density estimation method is adopted to obtain a wind power interval prediction result, and the wind power interval prediction result is verified. According to the method, the problems of poor decomposition and influence of different bandwidths of each component on an estimation error are effectively avoided, and the prediction precision of the wind power can be effectively improved.

Description

Technical Field

[0001] The present invention relates to the technical field of wind power prediction, and in particular to a wind power probability prediction method. Background Art

[0002] The world is practicing the concept of green development, aiming at clean and low-carbon development, and vigorously developing renewable energy represented by wind power. However, the intermittent and volatile nature of wind power places higher demands on the flexibility of the power grid. Due to the growing global demand for electricity and the high penetration rate of wind power, accurate wind farm wind power forecasting is an effective way to quantify wind energy uncertainty and ensure stable dispatch of wind energy resources.

[0003] Wind power prediction can be divided into two categories: deterministic prediction and uncertain prediction according to the different forms of the final prediction results. Deterministic prediction methods can be further divided into single model and combined model prediction. For example, Tuerxun W et al. used the LSTM single model to predict wind power. However, due to the strong volatility of wind energy, it is difficult for a single model to effectively learn the relationship between features. The combined model overcomes the weaknesses of the single model and becomes the main method for deterministic prediction. Hanifi et al. used the CNN-LSTM model to predict wind power, which can effectively process long-time series data and further improve the prediction accuracy. However, the CNN-LSTM model is overly dependent on gradient information, which may lead to gradient disappearance or explosion problems. The Stacking model can integrate multiple base models, take advantage of their strengths and overcome their weaknesses, and solve the gradient explosion problem. Qu et al. used the D-Stacking model to predict the power of wind farm clusters based on the different amplitudes of the decomposed subsequences, achieving higher prediction accuracy than combined models such as CNN-LSTM, but did not consider the impact of different amplitudes on the prediction results. Kong et al. used the attention mechanism method to focus on the main features, thereby improving the accuracy of the model. The prediction method combining attention mechanism and Multi-Stacking (FAMM-Stacking) has rarely been reported in the deterministic prediction of wind farm cluster power, which deserves further study. However, deterministic prediction is difficult to characterize the uncertainty of wind power, while probabilistic prediction is a powerful tool to quantify the deviation of prediction results caused by wind power uncertainty. While giving deterministic prediction results, it depicts the probability interval of wind power fluctuations, which is more conducive to the reasonable arrangement of spare capacity and thermal power unit startup plan.

[0004] There are many methods for wind power probability prediction. According to whether it depends on a specific type of probability distribution form, wind power probability prediction can be divided into parametric prediction and non-parametric prediction. Parametric probability prediction requires the assumption that wind power or prediction error distribution is Gaussian distribution, Weibull distribution, etc. For example, Shirzadi N and other scholars use Weibull distribution to simulate its probability distribution. However, wind energy has strong volatility and randomness, and the prediction error will change with the change of prediction conditions. The parameter distribution assumption is difficult to characterize the true probability distribution of the prediction object. Non-parametric probability prediction is a probability prediction method that does not give any distribution shape assumptions, which effectively avoids the problem of parametric distribution probability prediction. Common non-parametric probability prediction methods include quantile regression and kernel density estimation. Some scholars use adaptive quantile regression algorithms to predict the quantiles of wind power output, but multiple quantiles need to be modeled and calculated separately, which has low calculation efficiency. Yang and other scholars proposed a non-parametric kernel density estimation (KDE) method to improve the calculation efficiency and realize the modeling of the probability distribution of wind power, but the choice of bandwidth parameters has a great influence on the estimation results. Some researchers have used particle swarm optimization (PSO) and grid search (GS) to optimize KDE, achieving better prediction results than the unoptimized KDE method. However, the optimized KDE is a probabilistic prediction of the overall deterministic prediction results after superposition, while ignoring the impact of different bandwidths of each component on the estimation error. In order to minimize the estimation error as much as possible, a distributed KDE is proposed, which fits the error distribution of the deterministic prediction results by combining different components with the attention mechanism to obtain the final probabilistic prediction.

[0005] In order to further improve the accuracy of wind power prediction, scholars generally consider two aspects: time series decomposition and feature extraction. In terms of time series decomposition, there are many decomposition methods such as EMD, ICEEMDAN, and VMD. Nahid uses EMD decomposition and combines it with the CNN-LSTM model to obtain wind power prediction results. However, the signal decomposed by EMD has modal aliasing. Emeksiz and other scholars use ICEEMDAN decomposition to suppress modal aliasing to a certain extent, and the reconstruction error is relatively small, which can obtain better prediction results. Hu et al. used VMD to decompose the historical wind power time series, which effectively suppressed the modal aliasing of the signal and made the wind power prediction more accurate. However, the high-frequency strong non-stationary components generated by VMD decomposition will also lead to large prediction errors. At the same time, VMD punishes internal jumps based on the L2 smoothing stage transition, and does not handle burst signals well. Sibtain and other scholars combined the VMD-ICEEMDAN secondary decomposition technology with the LSTM model for prediction, which solved the problem of large prediction errors of high-frequency non-stationary components generated by VMD decomposition, but still did not solve the problem of poor decomposition of burst signals. Parri and Teeparthi used successive variational mode decomposition (SVMD) to describe the different modes of the signal by constructing a variational model, and solved the model through an optimization algorithm to extract the mode, which is more flexible and adaptable to more complex burst signals. Therefore, the rarely reported SVMD-ICEEMDAN secondary decomposition method is proposed to effectively solve the problems of poor prediction of high-frequency signals and poor decomposition of burst signals of complex time series. However, after the secondary decomposition of SVMD-ICEEMDAN, there are a large number of subsequences, which increases the time complexity of the model. Some researchers use permutation entropy and sample entropy to measure the complexity of subsequences and aggregate subsequences. However, the information entropy of a single scale is difficult to reflect the complex changes of each subsequence after wind power decomposition. To this end, the refined composite multi-scale weighted permutation entropy (RCMWPE) is designed to measure subsequences, so as to more reasonably aggregate subsequences. Based on this, it is proposed that SVMD-ICEEMDAN decomposition combined with the RCMWPE method (SIR method) is used to process wind power data.

[0006] In terms of feature extraction, some researchers have analyzed the spatiotemporal correlation of wind farm clusters based on directed graph networks and traditional correlation analysis. However, traditional correlations, such as Pearson correlation, are not friendly to extreme value correlations. Some scholars use the Copula structure to better simulate and display the extreme value data correlation in the original data, analyze the spatiotemporal characteristics of wind farms, and achieve better performance than traditional correlation analysis. However, graph networks mainly analyze short-range spatial relationships, lack spatial analysis between regions, and have omissions of spatiotemporal associations that have an important impact on prediction. The Moran Index is an effective tool for analyzing spatial relationships between regions. For example, Qu Zhijian and other scholars used the Moran Index to analyze the relationship between temperature, air pressure, and irradiance in Xinjiang, China, to predict solar irradiance.

[0007] In summary, in the existing technology, the prediction effect of high-frequency subsequence signals decomposed once is poor and the burst signal decomposition of complex time series is not good. The different bandwidths of each component will affect the estimation error. Secondly, it is impossible to accurately measure the complexity of each component, which is not conducive to improving the accuracy of wind power prediction. Summary of the invention

[0008] Based on this, the purpose of the present invention is to provide a wind power probability prediction method to solve the deficiencies in the above-mentioned prior art.

[0009] The present invention provides a method for wind power probability prediction, the method comprising: Collect geographical data, weather characteristics data and historical characteristics data of the wind farm cluster area; Calculate the Moran index value of the area where the wind farm cluster is located based on the geographic data, the weather characteristic data and the historical characteristic data of the wind farm, extract the spatial correlation characteristics between the areas where the wind farm cluster is located, and further extract the spatial characteristics in combination with the Copula graph network model, and form a final input matrix according to the Moran index value and the spatial characteristics; Decomposing the wind power time series data of the wind farm historical characteristic data by successive variational mode to obtain preliminary components, and decomposing the preliminary components by adaptive noise complete set empirical mode to obtain several components; The entropy values ​​of the components are calculated by using a fine composite multi-scale weighted permutation entropy, and the complexity of the components is comprehensively measured and analyzed by using a mean value method to obtain a plurality of comprehensive components; Inputting the final input matrix and several of the integrated components into the FAAM-Stacking model to obtain a final deterministic prediction result and verify the performance; A distributed kernel density estimation method is used according to the error of the final deterministic prediction result to obtain a wind power interval prediction result, and the wind power interval prediction result is verified.

[0010] Compared with the prior art, the beneficial effects of the present invention are as follows: through the final input matrix composed of Moran's index value and spatial characteristics, the geographical, weather and electric field historical characteristics of the area where the wind farm cluster is located can be effectively analyzed and extracted, and secondary decomposition is performed by combining successive variational mode decomposition and adaptive noise complete set empirical mode decomposition, and aggregation is performed through fine composite multi-scale weighted permutation entropy, which effectively avoids the problems of poor decomposition and different bandwidths of each component affecting the estimation error, and the final deterministic prediction result is predicted by the FAAM-Stacking model, and the wind power interval prediction result is obtained by the distributed kernel density estimation method, thereby effectively improving the prediction accuracy of wind power.

[0011] Furthermore, the step of calculating the Moran's index value of the area where the wind farm cluster is located based on the geographic data, the weather characteristic data and the historical characteristic data of the wind farm, extracting the spatial correlation characteristics between the areas where the wind farm cluster is located, and further extracting the spatial characteristics in combination with the Copula graph network model, and forming the final input matrix according to the Moran's index value and the spatial characteristics includes: The Moran index value of the area where the wind farm cluster is located is calculated based on the geographic data, the weather characteristic data and the historical characteristic data of the wind farm to screen out preliminary spatiotemporal characteristics, wherein the calculation expression of the Moran index value is: ; ; In the formula, represents the Moran's index value, Indicates the total number, Representation elements i and elements j The spatial weight between Indicates i The deviation of a value from the mean, Indicates j The deviation of a value from its mean, represents the aggregation of all spatial weights; Establishing a Copula graph network model, using the positional relationship of the adjacent wind farm clusters to form a graph network, screening out the nodes constituting the graph network through Copula theory and extracting relevant spatiotemporal features; The preliminary spatiotemporal features and the related spatiotemporal features are combined to obtain a new feature matrix, and the feature matrix is ​​subjected to mutual information dimensionality reduction to obtain a final input matrix.

[0012] Furthermore, the step of using successive variational mode decomposition to decompose the electric power time series data of the historical characteristic data of the wind farm, and using adaptive noise complete set empirical mode decomposition to decompose the preliminary components to obtain a plurality of components includes: Decomposing the wind power time series signal in the wind power time series into L-order modes and residual signals; Performing a minimization constraint on the L-order mode, and minimizing the residual signal using a first preset filter; Using a second preset filter to establish a final constraint condition, and constraining the L-order mode based on the final constraint condition to obtain a preliminary component; Adding Gaussian white noise to the high-frequency subsequence signal of the successive variational mode decomposition to obtain a first residual value; Defining a plurality of modes generated by the adaptive noise complete set empirical mode decomposition, and calculating the local mean of the first residual value by EMD plus Gaussian white noise to obtain a second residual value; The calculation of all the modes is repeated and the remaining residuals are made to satisfy monotonicity to obtain several components.

[0013] Furthermore, the wind power time series signal is decomposed into L-order modes and residual signals as follows: ; In the formula, represents the wind power time series signal, represents the L-order mode, represents the residual signal.

[0014] Furthermore, the step of using fine composite multi-scale weighted permutation entropy to calculate the entropy values ​​of the components, and using the mean method to comprehensively measure and analyze the complexity of the components to obtain the multiple comprehensive components includes: Averaging the electric power time series data to construct a plurality of coarse-grained time series; Reconstructing the phase space of the plurality of coarse-grained time series to obtain a reconstructed time series; Arranging the reconstructed time series in ascending order to obtain an arranged time series; Calculating weight values ​​of a plurality of subsequences in the arranged time series; Calculating a weighted probability value of the arranged time series in each arrangement mode based on the weight values ​​of the plurality of subsequences, and calculating an average frequency of occurrence of the weighted probability value in the arranged time series in each arrangement mode; Calculating the refined composite multi-scale weighted permutation entropy of the permuted time series to obtain multiple entropy values ​​of several components; The mean entropy of the components is obtained based on the mean method and the plurality of entropy values.

[0015] Furthermore, the expression for constructing several coarse-grained time series is: ; In the formula, represents the coarse-grained time series, represents the scale factor, Indicates The scale factor, represents a positive integer, for Integers in It means [1, ], Represents a discrete time series.

[0016] Furthermore, the step of inputting the final input matrix and the plurality of the comprehensive components into the FAAM-Stacking model to obtain a final deterministic prediction result and verify the performance includes: Inputting the comprehensive components into a multi-model fusion Multi-Stacking model according to different amplitudes to obtain a number of prediction values; Inputting the plurality of predicted values ​​into a feed-back attention mechanism to obtain a final deterministic prediction result; The standard root mean square error, mean absolute percentage error and determination coefficient are used to verify the final prediction results.

[0017] Furthermore, the expression for verifying the final prediction result is: ; ; ; In the formula, represents the standard root mean square error, represents the mean absolute percentage error, represents the coefficient of determination, It represents the difference between the maximum and minimum values ​​measured. Indicates the total number, Indicates the measured value, represents the predicted value of wind power, represents the average value of the measured values, Represents a positive integer.

[0018] Furthermore, the step of verifying the wind power interval prediction result includes: The prediction interval coverage and the prediction interval average bandwidth are used to verify the wind power interval prediction results.

[0019] Furthermore, the expression of the verification is: ; ; In the formula, represents the coverage of the prediction interval, represents the average bandwidth of the prediction interval, Indicates the total number, Represents a Boolean quantity, represents the number of samples, Indicates the range of change of the predicted target value. , Respectively represent The upper boundary of the prediction interval, The lower boundary of the prediction interval. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 is a flow chart of a wind power probability prediction method in an embodiment of the present invention; Figure 2 Schematic diagram of the FAAM-Stacking prediction model in an embodiment of the present invention; Figure 3 This is a comparison chart of interval predictions of different models at a 90% confidence level in an embodiment of the present invention.

[0021] The following specific implementation manner will further illustrate the present invention in conjunction with the above-mentioned drawings. DETAILED DESCRIPTION

[0022] In order to facilitate the understanding of the present invention, the present invention will be described more fully below with reference to the relevant drawings. Several embodiments of the present invention are given in the drawings. However, the present invention can be implemented in many different forms and is not limited to the embodiments described herein. On the contrary, the purpose of providing these embodiments is to make the disclosure of the present invention more thorough and comprehensive.

[0023] It should be noted that when an element is referred to as being "fixed to" another element, it may be directly on the other element or there may be a central element. When an element is considered to be "connected to" another element, it may be directly connected to the other element or there may be a central element at the same time. The terms "vertical", "horizontal", "left", "right" and similar expressions used herein are for illustrative purposes only.

[0024] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art to which the present invention belongs. The terms used herein in the specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. The term "and / or" used herein includes any and all combinations of one or more of the related listed items.

[0025] See also Figure 1 , which shows a wind power probability prediction method in an embodiment of the present invention, and the method includes steps S1 to S6: S1, collects geographical data, weather characteristic data and historical characteristic data of the wind farm cluster in the area; It should be noted that the geographical data of the area where the electric field is located and the historical characteristic data of the wind farm are collected, including the detailed longitude and latitude information of the regional boundary, weather data and historical data of wind power. In this embodiment, the electric field cluster data of a certain region in China are selected for analysis, including the longitude and latitude information of the boundaries of each county in the region, detailed historical weather information of 80 wind farms, and the alternative characteristics of each wind farm include wind speed, wind direction, wind speed and wind direction at the hub height, temperature, air pressure, air humidity and historical power generation of the wind turbine.

[0026] It is worth noting that the use of ArcGIS to collect geographic data of various districts and counties in a region can clearly distinguish the boundaries of various districts and counties; Collect data on wind farm clusters, the number of wind farms , weather and historical power data for each wind farm ,in, Represents the characteristic data dimension of the wind farm, Indicates temperature characteristics and wind speed characteristics.

[0027] S2, based on the geographic data, the weather characteristic data and the historical characteristic data of the wind farm, calculate the Moran index value of the area where the wind farm cluster is located, extract the spatial correlation characteristics between the areas where the wind farm cluster is located, and further extract the spatial characteristics in combination with the Copula graph network model, and form a final input matrix according to the Moran index value and the spatial characteristics; Specifically, the step S2 includes steps S21 to S23: S21, calculating the Moran index value of the area where the wind farm cluster is located based on the geographic data, the weather characteristic data and the historical characteristic data of the wind farm, so as to screen out preliminary spatiotemporal characteristics, wherein the calculation expression of the Moran index value is: ; ; In the formula, represents the Moran's index value, Indicates the total number, Representation elements i and elements j The spatial weight between Indicates i The deviation of a value from the mean, Indicates j The deviation of a value from its mean, represents the aggregation of all spatial weights; It can be understood that the Moran index values ​​between the districts and counties in the region are calculated to preliminarily screen out the preliminary spatiotemporal characteristics.

[0028] S22, establishing a Copula graph network model, using the positional relationship of the adjacent wind farm clusters to form a graph network, screening out the nodes constituting the graph network through Copula theory and extracting relevant spatiotemporal features, wherein the mathematical expression of Copula theory is as follows: Assumptions are n random variables, and their marginal distributions are Their joint distribution is , then there exists a function that “connects” the marginal distribution and the joint distribution

[0029]

[0030] According to the inverse transformation of the CDF of the marginal distribution, that is, , we can get the expression of Copula function: .

[0031] It can be understood that by establishing a Copula graph network model, the positional relationship between adjacent wind farm nodes is used to form a possible graph network, and then the Copula theory is used to screen out suitable nodes to form the graph network and extract relevant spatiotemporal features.

[0032] S23, combining the preliminary spatiotemporal features and the related spatiotemporal features to obtain a new feature matrix, and performing mutual information dimensionality reduction on the feature matrix to obtain a final input matrix.

[0033] S3, using successive variational mode decomposition of the wind power time series data of the wind farm historical characteristic data to obtain preliminary components, and using adaptive noise complete set empirical mode decomposition of the preliminary components to obtain several components; Specifically, the step S3 includes steps S31 to S36: S31, decomposing a wind power time series signal in the wind power time series into L-order modes and a residual signal; It can be understood that the wind power time series signal is decomposed into two parts: the L-order mode and the residual signal. The expression of decomposing the wind power time series signal into the L-order mode and the residual signal is: ; In the formula, represents the wind power time series signal, represents the L-order mode, represents the residual signal; The residual signal includes the unprocessed part And the sum of the modes obtained before , if you want to satisfy , then L-order mode constraints are required.

[0034] S32, performing minimization constraint on the L-order mode, and using a first preset filter to minimize the residual signal; It can be understood that the L-order mode realizes the minimization constraint, and the expression of the constraint condition is: ; In the formula, represents the convolution operation, represents the center frequency of the L-order mode, The code for the constraint expression is one. represents the partial derivative with respect to time t, represents the Dirac function, represents an imaginary number, represents the base of natural logarithms, Indicates time; On the basis that the L-order mode has effective components, the residual signal should be minimized as much as possible. In order to ensure that this constraint can be stably implemented, a suitable filter should be selected. The frequency response expression is: ; The constraint that should be established is expressed as: ; In the formula, represents the frequency response, Indicates frequency, represents the equilibrium parameter, The code 2 represents the constraint expression. Represents a filter.

[0035] S33, using a second preset filter to establish a final constraint condition, and constraining the L-order mode based on the final constraint condition to obtain a preliminary component; It is understandable that after the above two constraints, it is impossible to distinguish the L-order mode and the L-1 mode. Based on the idea of ​​constraints, a suitable filter is selected and the frequency response expression is: ; In the formula, Expressed as the impulse response to the filter, Represents an integer; Thus, the constraint can be established, and the expression is:

[0036] In the formula, The code three represents the constraint expression. Indicates i An impulse response; Therefore, the final constraint is to ensure that the signal can be completely reconstructed, expressed as:

[0037] In the formula, represents the Lth-order mode, Represents the unprocessed part of the signal; Therefore, the problem of extracting modal components can be considered as a problem of minimizing three constraints, expressed as: ; In the formula, For balance , , ; After successive variational mode decomposition, the original wind power data is decomposed into 6 subsequences; S34, adding Gaussian white noise to the high-frequency subsequence signal of the successive variational mode decomposition to obtain a first residual value; It needs to be explained that the high-frequency components of the successive variational mode decomposition still have problems of instability and poor prediction effect. However, by using the adaptive noise complete set empirical mode decomposition algorithm through the local mean method, the problems of modal distortion and instability can be effectively solved. Specifically: Add Gaussian white noise to the high-frequency subsequence signal of successive variational mode decomposition. The expression is: ; In the formula, represents the signal after adding Gaussian white noise, represents the high-frequency subsequence signal of successive variational mode decomposition, represents the standard deviation of the added Gaussian white noise, represents Gaussian white noise with zero mean and variance, is the function for calculating IMF1, where the expression for the first residual value is:

[0038] In the formula, represents the first residual value, Represents the high-frequency sub-signal after a decomposition The total number of data, Indicates calculation of IMF function.

[0039] S35, defining a plurality of modes generated by the adaptive noise complete set empirical mode decomposition, and calculating the local mean of the first residual value after adding Gaussian white noise through EMD to obtain a second residual value; It should be explained that K is defined as the kth mode generated by the adaptive noise complete set empirical mode decomposition. When k=1, the first mode is obtained, and the expression is:

[0040] In the formula, Indicates the first mode; When k=2, the EMD plus Gaussian white noise is used to calculate The local mean of , we get the second residual, which is expressed as:

[0041] In the formula, represents the second residual value, represents the standard deviation of the added Gaussian white noise, Indicates the calculation of the second IMF function; when Calculate the kth residual and kth mode, the expression is:

[0042]

[0043] In the formula, represents the Kth residual, represents the K-1th residual, represents the standard deviation of the Gaussian white noise added to C_(k-1), Indicates the calculation of the Kth IMF function, represents the Kth IMF value.

[0044] S36, repeatedly calculating all the modes, and making the remaining residuals satisfy monotonicity, so as to obtain a plurality of components; It is understandable that the calculation is repeated for all IMFs until the remaining residual It satisfies monotonicity and cannot be decomposed any further. Finally, the original signal is decomposed into two parts, expressed as:

[0045] In the formula, represents the original signal, Indicates i IMF values, and finally, the adaptive noise complete set empirical mode decomposes the highest frequency component into 14 new components again, plus the remaining components of the successive variational mode decomposition, a total of 19 subsequences are obtained.

[0046] S4, using a refined composite multi-scale weighted permutation entropy to calculate the entropy values ​​of the several components, and using a mean method to comprehensively measure and analyze the complexity of the several components to obtain several comprehensive components; Specifically, the step S4 includes steps S41 to S47: S41, averaging the electric power time series data to construct a plurality of coarse-grained time series; It should be noted that 19 components are obtained after the secondary decomposition, which increases the time complexity. Therefore, it is considered to use the information entropy to measure the complexity of the subsequence and aggregate the subsequence to reduce the time complexity. Specifically, the electric power time series data in the window is first averaged to construct several coarse-grained time series. The expression for constructing several coarse-grained time series is: ; In the formula, represents the coarse-grained time series, represents the scale factor, Indicates The scale factor, represents a positive integer, for Integers in It means [1, ], Represents a discrete time series.

[0047] S42, performing phase space reconstruction on the plurality of coarse-grained time series to obtain a reconstructed time series; It can be understood that the expression for phase space reconstruction of coarse-grained time series is: ; In the formula, represents a coarse-grained time series, represents the reconstructed time series, ,⋯, Respectively expressed in The coarse-grained time series under The coarse-grained time series under represents the delay factor, represents the embedding dimension.

[0048] S43, arranging the reconstructed time series in ascending order to obtain an arranged time series; It can be understood that the expression of the arranged time series is: ; In the formula, , , , Respectively expressed in The coarse-grained time series after sorting, The coarse-grained time series after sorting, The sorted coarse-grained time series, where Indicates the number of embeddings.

[0049] S44, calculating weight values ​​of several subsequences in the arranged time series; The calculation expression of the weight value of each subsequence is: ; In the formula, express A coarse-grained time series, represents the mean of the time series, where The calculation expression is: .

[0050] S45, calculating a weighted probability value of the arranged time series in each arrangement mode based on the weight values ​​of the plurality of subsequences, and calculating an average frequency of occurrence of the weighted probability value in the arranged time series in each arrangement mode; It is understandable that any subsequence The feature information is used with weight values and arrangement mode For this time series, there are k kinds of arrangement modes, and each arrangement mode The expression of the weighted probability value is: ; in, , , , Represent the first reconstructed component, the second reconstructed component, the The reconstruction component, express The factorial of Indicates that the coarse-grained time series is in the arrangement mode Under the effect The average frequency of occurrence.

[0051] S46, calculating the fine composite multi-scale weighted permutation entropy of the permuted time series to obtain multiple entropy values ​​of several components; It can be understood that the expression for calculating the fine composite multi-scale weighted permutation entropy of the permuted time series is:

[0052] In the formula, represents the refined composite multi-scale weighted permutation entropy, represents the time series after arrangement, express The average frequency.

[0053] S47, obtaining mean entropy of the components based on the mean method and the multiple entropy values; It can be understood that the expression of the mean entropy of several components is:

[0054] AVR stands for mean entropy, represents the mean of the fine composite multi-scale weighted permutation entropy, represents the fine composite multi-scale weighted permutation entropy. In this embodiment, the components are reconstructed according to the similar mean entropy, and finally four wind power subsequences with different amplitudes are obtained. .

[0055] S5, inputting the final input matrix and the plurality of the comprehensive components into the FAAM-Stacking model to obtain a final deterministic prediction result and verify the performance; Specifically, the step S5 includes steps S51 to S53: S51, inputting the comprehensive component into a multi-model fusion Multi-Stacking model according to different amplitudes to obtain a number of prediction values; It is understandable that the aggregated wind power subsequence Analyze the amplitude and input it into the Multi-Stacking model of multi-model fusion according to the amplitude, such as Figure 2 As shown, the predicted value is ; S52, inputting the plurality of predicted values ​​into a feed-back attention mechanism to obtain a final deterministic prediction result; It is understandable that the predicted value Input into the feedback attention mechanism to obtain the final deterministic prediction result, which is expressed as: ; In the formula, Represents the predicted result, represents the first weight, represents the second weight, represents the third weight, represents the first amplitude, represents the second amplitude, represents the third amplitude; It is worth mentioning that the Multi-Stacking model with feedback attention mechanism is the FAAM-Stacking model.

[0056] S53, using standard root mean square error, mean absolute percentage error and determination coefficient to verify the final prediction result; It can be understood that the expression for verifying the final prediction result is: ; ; ; In the formula, represents the standard root mean square error, represents the mean absolute percentage error, represents the coefficient of determination, It represents the difference between the maximum and minimum values ​​measured. Indicates the total number, Indicates the measured value, represents the predicted value of wind power, represents the average value of the measured values, Represents an integer.

[0057] S6, obtaining a wind power interval prediction result according to the error of the final deterministic prediction result and using a distributed kernel density estimation method, and verifying the wind power interval prediction result; It is understandable that in the specific implementation, the calculation formula of the kernel density estimation is first clarified, and the calculation formula of the kernel density estimation is:

[0058] In the formula, represents the estimated probability density function, represents the kernel function, Indicates taking the absolute value; represents the indicative function, Represents a random event, n represents the number of error samples, Indicates the window width.

[0059] It should be noted that the distributed kernel density estimation method is based on the kernel density estimation. For the prediction results of each component, the probability density function is given different attention, thereby reducing the error and achieving the purpose of improving wind power prediction. The expression of the distributed kernel density estimation method is:

[0060]

[0061] In the formula, Indicates the window width, Indicates prediction error samples, express The weight coefficient of Indicates bandwidth is The kernel density expression is, is the expression for the distributed kernel density, and Q is the total bandwidth. represents the sth window width, Represents an integer; The expression of the verification is: ; ; In the formula, represents the coverage of the prediction interval, represents the average bandwidth of the prediction interval, Indicates the total number, Represents a Boolean quantity, represents the number of samples, Indicates the range of change of the predicted target value. , Respectively represent The upper boundary of the prediction interval, The lower boundary of the prediction interval, is the confidence level, The smaller it is, the better the sensitivity of the prediction interval.

[0062] It should be noted that the prediction error is obtained by the above evaluation index formula to obtain a specific evaluation result value, which illustrates the effectiveness of the present invention. The present invention is compared with its four benchmark models: LSTM, CNN-LSTM, and CNN-BiGRU. The example of a wind farm cluster in a certain autonomous region in China is used for analysis. The comparison results are shown in Figure 2. Figure 3 As shown in the figure, at a confidence level of 90%, the prediction interval of the proposed method is the smallest, indicating that the effect of the present invention is the best.

[0063] In summary, the wind power probability prediction method in the above-mentioned embodiment of the present invention can effectively analyze and extract the geographical, weather and electric field history characteristics of the area where the wind farm cluster is located through the final input matrix composed of Moran index values ​​and spatial features, and perform secondary decomposition by combining successive variational mode decomposition and adaptive noise complete set empirical mode decomposition, and aggregate through fine composite multi-scale weighted permutation entropy, which effectively avoids the problems of poor decomposition and different bandwidths of each component affecting the estimation error. The final deterministic prediction result is predicted by the Multi-Stacking model with a feedback attention mechanism, and the wind power interval prediction result is obtained by the distributed kernel density estimation method, thereby effectively improving the prediction accuracy of wind power.

[0064] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "examples", "specific examples", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

[0065] The above-mentioned embodiments only express several implementation methods of the present invention, and the description thereof is relatively specific and detailed, but it cannot be understood as limiting the scope of the patent of the present invention. It should be pointed out that, for ordinary technicians in this field, several variations and improvements can be made without departing from the concept of the present invention, which all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be subject to the attached claims.

Claims

1. A wind power probability prediction method, characterized in that: The method comprises: Collect geographical data, weather characteristics data and historical characteristics data of the wind farm cluster area; Calculate the Moran index value of the area where the wind farm cluster is located based on the geographic data, the weather characteristic data and the historical characteristic data of the wind farm, extract the spatial correlation characteristics between the areas where the wind farm cluster is located, and further extract the spatial characteristics in combination with the Copula graph network model, and form a final input matrix according to the Moran index value and the spatial characteristics; Decomposing the wind power time series data of the wind farm historical characteristic data by successive variational mode to obtain preliminary components, and decomposing the preliminary components by adaptive noise complete set empirical mode to obtain several components; The entropy values ​​of the components are calculated by using a fine composite multi-scale weighted permutation entropy, and the complexity of the components is comprehensively measured and analyzed by using a mean value method to obtain a plurality of comprehensive components; Inputting the final input matrix and several of the integrated components into the FAAM-Stacking model to obtain a final deterministic prediction result and verify the performance; A distributed kernel density estimation method is used according to the error of the final deterministic prediction result to obtain a wind power interval prediction result, and the wind power interval prediction result is verified.

2. The wind power probability prediction method according to claim 1, characterized in that: The step of calculating the Moran's index value of the area where the wind farm cluster is located based on the geographic data, the weather characteristic data and the historical characteristic data of the wind farm, extracting the spatial correlation features between the areas where the wind farm cluster is located, and further extracting the spatial features in combination with the Copula graph network model, and forming the final input matrix according to the Moran's index value and the spatial features includes: The Moran index value of the area where the wind farm cluster is located is calculated based on the geographic data, the weather characteristic data and the historical characteristic data of the wind farm to screen out preliminary spatiotemporal characteristics, wherein the calculation expression of the Moran index value is: ； ； In the formula, represents the Moran's index value, Indicates the total number, Representation elements i and elements j The spatial weight between Indicates i The deviation of a value from the mean, Indicates j The deviation of a value from its mean, represents the aggregation of all spatial weights; Establishing a Copula graph network model, using the positional relationship of the adjacent wind farm clusters to form a graph network, screening out the nodes constituting the graph network through Copula theory and extracting relevant spatiotemporal features; The preliminary spatiotemporal features and the related spatiotemporal features are combined to obtain a new feature matrix, and the feature matrix is ​​subjected to mutual information dimensionality reduction to obtain a final input matrix.

3. The wind power probability prediction method according to claim 1, characterized in that: The step of using successive variational mode decomposition of the electric power time series data of the historical characteristic data of the wind farm, and using adaptive noise complete set empirical mode decomposition of the preliminary components to obtain a plurality of components includes: Decomposing the wind power time series signal in the wind power time series into L-order modes and residual signals; Performing a minimization constraint on the L-order mode, and minimizing the residual signal using a first preset filter; Using a second preset filter to establish a final constraint condition, and constraining the L-order mode based on the final constraint condition to obtain a preliminary component; Adding Gaussian white noise to the high-frequency subsequence signal of the successive variational mode decomposition to obtain a first residual value; Defining a plurality of modes generated by the adaptive noise complete set empirical mode decomposition, and calculating the local mean of the first residual value by EMD plus Gaussian white noise to obtain a second residual value; The calculation of all the modes is repeated and the remaining residuals are made to satisfy monotonicity to obtain several components.

4. The wind power probability prediction method according to claim 3 is characterized in that: The wind power time series signal is decomposed into L-order modes and residual signals as follows: ； In the formula, represents the wind power time series signal, represents the L-order mode, represents the residual signal.

5. The wind power probability prediction method according to claim 1, characterized in that: The step of using fine composite multi-scale weighted permutation entropy to calculate the entropy values ​​of the components and using the mean method to comprehensively measure and analyze the complexity of the components to obtain the multiple comprehensive components includes: Averaging the electric power time series data to construct a plurality of coarse-grained time series; Reconstructing the phase space of the plurality of coarse-grained time series to obtain a reconstructed time series; Arranging the reconstructed time series in ascending order to obtain an arranged time series; Calculating weight values ​​of a plurality of subsequences in the arranged time series; Calculating a weighted probability value of the arranged time series in each arrangement mode based on the weight values ​​of the plurality of subsequences, and calculating an average frequency of occurrence of the weighted probability value in the arranged time series in each arrangement mode; Calculating the refined composite multi-scale weighted permutation entropy of the permuted time series to obtain multiple entropy values ​​of several components; The mean entropy of the components is obtained based on the mean method and the plurality of entropy values.

6. The wind power probability prediction method according to claim 5, characterized in that: The expression for constructing several coarse-grained time series is: ； In the formula, represents the coarse-grained time series, represents the scale factor, Indicates The scale factor, represents a positive integer, for Integers in It means [1, ], Represents a discrete time series.

7. The wind power probability prediction method according to claim 1, characterized in that: The step of inputting the final input matrix and the plurality of the comprehensive components into the FAAM-Stacking model to obtain the final deterministic prediction result and verify the performance comprises: Inputting the comprehensive components into a multi-model fusion Multi-Stacking model according to different amplitudes to obtain a number of prediction values; Inputting the plurality of predicted values ​​into a feed-back attention mechanism to obtain a final deterministic prediction result; The standard root mean square error, mean absolute percentage error and determination coefficient are used to verify the final prediction results.

8. The wind power probability prediction method according to claim 7, characterized in that: The expression for verifying the final prediction result is: ； ； ； In the formula, represents the standard root mean square error, represents the mean absolute percentage error, represents the coefficient of determination, It represents the difference between the maximum and minimum values ​​measured. Indicates the total number, Indicates the measured value, represents the predicted value of wind power, represents the average value of the measured values, Represents a positive integer.

9. The wind power probability prediction method according to claim 1, characterized in that: The step of verifying the wind power interval prediction result comprises: The prediction interval coverage and the prediction interval average bandwidth are used to verify the wind power interval prediction results.

10. The wind power probability prediction method according to claim 9, characterized in that: The expression of the verification is: ； ； In the formula, represents the coverage of the prediction interval, represents the average bandwidth of the prediction interval, Indicates the total number, Represents a Boolean quantity, represents the number of samples, Indicates the range of change of the predicted target value. , Respectively represent The upper boundary of the prediction interval, The lower boundary of the prediction interval.

Images