A method for day-ahead prediction of wind power cluster power

Abstract

The present invention relates to the technical field of short-term wind power forecasting, and discloses a method for day-ahead wind power cluster power forecasting. The method comprises: summarizing the spatiotemporal correlation characteristics of a wind power cluster into the long-distance convergence effect and the short-distance temporal causal relationship between wind farms; using an improved set pair analysis method, a simultaneity discrimination method, and a variance discrimination method to obtain a convergence effect adjacency matrix; using the Pearson correlation coefficient and the Granger causality test to obtain a temporal causal adjacency matrix; proposing an error evaluation characteristic coefficient formula for constructing an error adjacency matrix; and improving a traditional graph convolutional neural network to propose a graph neural network structure based on a multi-channel attention mechanism. The present invention proposes a graph convolution prediction method that considers the fusion of the wind power cluster convergence effect and the multi-channel attention mechanism, which has strong prediction stability, high prediction performance, effective prediction results, and strong practicality.

Description

Technical Field

[0001] The present invention relates to the technical field of short-term wind power prediction, and in particular to a method for day-ahead prediction of wind power cluster power. Background Art

[0002] In recent years, installed capacity of renewable energy has continued to increase rapidly, accelerating the transformation of the global energy structure. Due to the randomness, intermittency, and volatility of wind power, as its scale continues to expand and its share of the energy mix increases, its impact on the safe and stable operation of power systems will also grow. To ensure regional power balance, large-scale wind power cluster forecasting is imperative.

[0003] A day-ahead wind power forecast is a prediction for the next day, starting from the prediction time, with a time resolution of 15 minutes. The significance of a day-ahead wind power forecast is that it provides a reference for planning wind farm maintenance and backup capacity.

[0004] There are many existing studies on cluster wind power prediction, but there are still some shortcomings: there are studies on the output evolution laws between wind farms over long periods of time and wide areas, but they are rarely applied; causal algorithms and correlation algorithms have difficulty representing the spatiotemporal correlations between wind farms that are far apart; the description of the spatiotemporal correlations of wind power clusters is vague, and there is no clear definition or summary; the application of errors is limited to the correction of prediction results and probabilistic prediction, and there is a lack of research on the extraction and application of error characteristics. Summary of the Invention

[0005] The purpose of the present invention is to provide a method for predicting the power of a wind power cluster day-ahead, which is based on the spatiotemporal characteristics of the wind power cluster and provides a definition summary of the spatiotemporal correlation of the wind power cluster, providing a method for predicting the power of a wind power cluster day-ahead that is clear in meaning, scientific, reasonable and effective, reliable and stable in prediction, can take into account the spatiotemporal correlation and convergence effect of the wind power cluster, and at the same time incorporates the wind power prediction error characteristics.

[0006] In order to achieve the above purpose, the technical solutions adopted are as follows:

[0007] The present invention provides a method for predicting wind power cluster power day-ahead, the method comprising:

[0008] Based on the wind power sequence, construct the adjacency matrix A s , simultaneous rate adjacency matrix A t , variance adjacency matrix A v ;

[0009] According to the adjacency matrix A s , simultaneous rate adjacency matrix A t and variance adjacency matrix A v , construct the convergence effect adjacency matrix A a;

[0010] Use Pearson correlation coefficient to construct the adjacency matrix A r , and construct the causal adjacency matrix A based on Granger causality check g ; Wherein, the adjacency matrix A r and the causal adjacency matrix A g Used to characterize the relationship between closely spaced wind farms;

[0011] Based on the adjacency matrix A r and the Granger causality adjacency matrix A g , construct the temporal causal adjacency matrix A b ;

[0012] Construct the error characteristic adjacency matrix A e ; Wherein, the error characteristic adjacency matrix A e Used to characterize the errors between wind farms;

[0013] A graph convolutional neural network model based on the channel attention mechanism is constructed. In the graph convolutional neural network model, multiple graph convolution modules are designed to input adjacency matrices with different attributes to form multiple channels. Each channel is used to extract different characteristic laws in the wind power cluster and obtain multiple features. The function definition is as follows:

[0014]

[0015] Where, The l+1th layer features representing the convergence effect, The l+1th layer features representing temporal causality, Represents the l+1th layer feature of the error, f represents the graph convolution function, The l-th layer feature representing the convergence effect, The l-th layer features representing temporal causality, The lth layer feature representing the error;

[0016] Based on multiple channels, the attention mechanism is introduced to learn the weight of each feature for prediction importance. Finally, multiple characteristics are combined to obtain the day-ahead wind power cluster power prediction result. The function is defined as follows:

[0017]

[0018] In the formula, Softmax(·) represents the normalization function, concat(·) represents the concatenation function, Output represents the day-ahead wind power cluster power forecast result, W a represents the weight of the convergence effect feature, W b Represents the weight of temporal causal features, W e represents the weight of the error feature, H represents the fusion feature, represents the learnable parameter of the l+1th layer convergence effect, represents the learnable parameters of the l+1th layer of temporal causality, Represents the learnable parameters of the error features of the l+1th layer.

[0019] Furthermore, the adjacency matrix A is constructed by the following method: s :

[0020] The following formula is used to traverse and calculate the connection degree between different wind farms in the wind power cluster to obtain the connection degree matrix:

[0021]

[0022]

[0023] Where, μ(A,B) is the connection matrix between wind power sequence A and wind power sequence B; {Δx i ,Δx i+1}、{Δx′ i ,Δx′ i+1} respectively represent the adjacent difference values ​​of the i-th identical characteristic element of wind power sequence A and wind power sequence B; {Δx j ,Δx j+1}、{Δx′ j ,Δx' j+1} respectively represent the adjacent difference values ​​of the jth opposite characteristic elements of wind power sequence A and wind power sequence B; I is the difference coefficient; J is the opposition identification coefficient; N is the total number of features in the set; α i is the characteristic strength of the i-th identical set pair; β j is the characteristic strength of the jth opposite set pair; S′ is the total strength of the same set pair; F′ is the total strength of the uncertainty set pair; P′ is the total strength of the opposite set pair;

[0024] In ascending order, select the first m elements in the connection matrix, where m is the average number of connections of the simultaneity rate and variance, retain m element values, and set the rest to zero to form a matrix M;

[0025] Use matrix M to construct the adjacency matrix A based on set pair analysis s , the calculation process is as follows:

[0026]

[0027] Where μ ij is the connection degree between wind farm No. i and No. j; ∞ means there is no connection between the two wind farms; τ is a constant used to avoid the edge weight being 0; S ij Represents the edge weights established by set pair analysis.

[0028] Furthermore, the simultaneous rate adjacency matrix A is constructed by the following method: t :

[0029] Traverse and calculate the daily synchronization rate of every two wind farms in the wind power cluster to form the matrix T k×i×j , the formula is as follows:

[0030]

[0031] Where, represents the power sequence of the i-th wind farm on the k-th day, represents the power sequence of wind farm No. j on day k, Cap i Cap represents the operating capacity of wind farm No. i, j represents the operating capacity of wind farm No. j; represents the simultaneous rate matrix of the wind power cluster on the kth day, represents the simultaneity rate of adding the power of wind farms No. i and No. j on day k;

[0032] For the matrix T k×i×j Discriminate each sub-matrix in the matrix to obtain the discriminant matrix of each day, and sum up the discriminant matrices of each day to obtain the simultaneous rate discriminant matrix H i×j , the formula is as follows:

[0033]

[0034] Where, represents the discriminant matrix of the kth day, represents the judgment result of row i and column j on day k, h ij Represents the sum of the judgment results of the i-th row and j-th column of all days; represents the simultaneous rate of power of wind farm No. j on day k;

[0035] Using the simultaneous rate discriminant matrix to calculate the adjacency matrix A of the directed weighted graph t The specific formula for constructing is as follows:

[0036]

[0037] Where W ij represents the weight of wind farm No. i pointing to wind farm No. j; t ij Represents the edge weight established by the simultaneous rate.

[0038] Furthermore, the variance adjacency matrix A is constructed by the following method: v :

[0039] Construct a variance matrix and construct a variance adjacency matrix A based on the variance matrix v ; Among them, the construction formula of the variance matrix is ​​as follows:

[0040]

[0041]

[0042] Where V k×i×j represents the variance matrix, represents the normalized sequence of the sum of wind farm No. i and wind farm No. j, represents the sequence mean, n is the number of elements in the wind power sequence, w is the sequence number of the wind power sequence, represents the variance matrix of the wind power cluster on day k, It represents the sum of the variance of wind farms No. i and No. j on day k.

[0043] Furthermore, according to the adjacency matrix A s , simultaneous rate adjacency matrix A t and variance adjacency matrix A v , the convergence effect adjacency matrix A is constructed by the following formula a :

[0044]

[0045] Where a ij represents the edge weight established by the convergence effect, s ij represents the edge weight constructed by the improved set pair analysis, t ij represents the edge weight constructed by the simultaneous rate, v ij Denotes the edge weights of the variance construction.

[0046] Furthermore, the adjacency matrix A is constructed using the Pearson correlation coefficient r ,include:

[0047] Traverse and calculate the daily synchronization rate of every two wind farms in the wind power cluster to form the matrix T k×i×j , the formula is as follows:

[0048]

[0049]

[0050] Where, represents the correlation coefficient matrix of the wind power cluster on day k; represents the correlation coefficient between the power of wind farm No. i and No. j on day k; denote the sequence of the kth day for wind farms i and j, respectively; represent the mean values ​​of wind farm No. i and No. j on the kth day respectively; express;

[0051] For the matrix T k×i×jDiscriminate each sub-matrix in the matrix to obtain the discriminant matrix of each day, and sum up the discriminant matrices of each day to obtain the simultaneous rate discriminant matrix H i×j , the formula is as follows:

[0052]

[0053] Where, represents the discriminant matrix of the kth day, represents the judgment result of row i and column j on day k, h ij Represents the sum of the judgment results of the i-th row and j-th column of all days;

[0054] Using the simultaneous rate discriminant matrix to calculate the adjacency matrix A of the directed weighted graph r 's construction.

[0055] Furthermore, the causal adjacency matrix A is constructed based on Granger causality check. g ,include:

[0056] The Granger test is used to test whether the daily power series of different wind farms have Granger causality, thereby obtaining a discriminant matrix, and constructing a causal adjacency matrix A based on the discriminant matrix. g Among them, the calculation formula for using the Granger test to test whether the daily power series of different wind farms have Granger causality is as follows:

[0057]

[0058] Where, represents the Granger causality of wind farm No. i to wind farm No. j in the kth day; It represents the Granger test value of wind farm No. i on wind farm No. j in the kth day; Fa represents the Granger test threshold.

[0059] Furthermore, construct the error characteristic adjacency matrix A e ,include:

[0060] The error characteristic adjacency matrix A is obtained based on the error smoothing evaluation coefficient. e ; Wherein, the error smoothing evaluation coefficient is expressed as:

[0061]

[0062] Where, represents the normalized value of the sum of the errors of wind farms No. i and No. j on day k; They represent the k-th day prediction error series of wind farm No. i and wind farm No. j respectively.

[0063] Furthermore, the method further comprises:

[0064] Select normalized root mean square error NRMSE, normalized mean absolute error NMAE and coefficient of determination R 2 As the evaluation index of the prediction model performance, the calculation formulas are:

[0065]

[0066] Where n is the length of the predicted power sequence, y i,pre and y i,ture are the predicted and actual values ​​of power, It represents the average value of actual power, and Cap is the installed capacity.

[0067] The beneficial effects of the present invention are:

[0068] This paper examines the spatiotemporal characteristics of wind farm clusters and proposes a feature extraction method that simultaneously accounts for both near- and long-range connections between wind farms. Furthermore, based on the error characteristics of wind power, a method is proposed that directly incorporates wind power prediction errors into the feature extraction and prediction process. Finally, a multi-channel attention mechanism for graph convolutional networks is proposed, which integrates adjacency matrix information from multiple different attributes to improve prediction accuracy and reliability. BRIEF DESCRIPTION OF THE DRAWINGS

[0069] Figure 1 A flow chart of a method for day-ahead power prediction of a wind power cluster according to an embodiment of the present invention is shown.

[0070] Figure 2 FIG. 4 shows a schematic diagram of set pair characteristic evaluation according to an embodiment of the present invention.

[0071] Figure 3 1 shows a characteristic strength analysis diagram according to an embodiment of the present invention; wherein (a) is a case where the opposite characteristic strength is the highest; (b) is a case where the opposite characteristic strength is relatively weak; (c) is a case where the opposite characteristic strength is the highest; (d) is a case where the opposite characteristic strength is the highest;

[0072] Figure 4 The figure shows a structural diagram of a graph convolutional neural network based on a channel attention mechanism according to an embodiment of the present invention;

[0073] Figure 5 Figures showing prediction results of cluster prediction strategies according to embodiments of the present invention; (a) comparison of prediction curves of different methods in cluster A; (b) comparison of prediction curves of different methods in cluster B; (c) comparison of prediction curves of different methods in cluster C;

[0074] Figure 6The figure shows the prediction error distribution diagram of various prediction strategies according to the embodiments of the present invention (the unit of frequency is "pieces"); among them, (a) the prediction error frequency statistics of different methods in cluster A; (b) the prediction error frequency statistics of different methods in cluster B; (c) the prediction error frequency statistics of different methods in cluster C. DETAILED DESCRIPTION

[0075] The following describes the embodiments of the present invention through specific examples. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments. The details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the following embodiments and features in the embodiments can be combined with each other unless they conflict.

[0076] The specific implementation of the present invention is further described in detail below with reference to the accompanying drawings and examples.

[0077] The embodiment of the present invention provides a method for predicting the power of a wind power cluster. Figure 1 As shown, the wind power cluster power day-ahead prediction method includes the following steps:

[0078] Step S10: Obtain the adjacency matrix A based on set pair analysis by using the improved set pair analysis method s .

[0079] Step S20: Obtain the simultaneous rate adjacency matrix A by calculating the simultaneous rate changes between each wind farm. t .

[0080] Step S30: Establish the variance adjacency matrix A v , the set pair analysis adjacency matrix A s , simultaneous rate adjacency matrix A t Take the intersection of the three and the variance adjacency matrix to get the convergence effect adjacency matrix A a .

[0081] Step S40: Obtain the correlation coefficient adjacency matrix A through the Pearson correlation coefficient r ;

[0082] Step S50: Use Granger causality test to obtain the causal adjacency matrix A g , the correlation coefficient adjacency matrix A r and the causal adjacency matrix A g Take the intersection to get the temporal causal adjacency matrix A b .

[0083] Step S60: Using the initial prediction error, construct the error characteristic adjacency matrix Ae .

[0084] Step S70: Input the NWP data of the wind power cluster, the convergence effect adjacency matrix, the temporal causal adjacency matrix, and the error characteristic adjacency matrix into a graph convolutional neural network based on a multi-channel attention mechanism to obtain predicted power.

[0085] The following embodiment of the present invention will describe in detail the specific implementation process of the above seven steps.

[0086] Regarding step S10, set pair analysis, also known as connection mathematics, is a new system analysis method that uses connection degree to deal with uncertainty problems. The core idea of ​​set pair analysis is to analyze and treat the deterministic and uncertain connections between the objective objects being studied as an uncertain system. Determinism includes the two aspects of "identity" and "opposition", while uncertainty refers solely to "difference". Set pair analysis analyzes objects and their systems through the three aspects of identity, difference, and opposition. Based on the relationship between sameness, difference, and opposition, the sameness, difference, and opposition connection degree between two sets is expressed as:

[0087]

[0088] Where: μ(A, B) is the degree of connection between set A and set B; N is the total number of features in the set, N = S + P + F; S is the number of elements in the same state (same characteristics) in the two sets; P is the number of elements in the opposite state (opposite characteristics) in the two sets; F is the number of elements in the different state (different characteristics) in the two sets, that is, the number of features in the sets that are neither identical nor opposite; I is the difference coefficient, which can be taken in the interval (-1, 1) according to the specific situation; J is the opposition identification coefficient.

[0089] For set pair analysis, determining the characteristics of each element in two sequences is crucial. In the set pair analysis method defined in this paper, the total number of characteristics, N+2, is the total number of elements in the sequence. That is, except for the first two elements in a sequence, all other elements in the sequence have the characteristics.

[0090] However, existing set pair analysis methods only count the number of characteristics of elements without considering the strength of the characteristics. Therefore, the improved set pair analysis adds a strength evaluation of the characteristics to the original set pair analysis to more reasonably characterize the strength of the smoothing effect between two series. The specific formula is as follows:

[0091]

[0092] Among them, {Δx i ,Δx i+1}, {Δx′ i ,Δx i '+1} respectively represent the adjacent difference values ​​of the i-th identical characteristic element of sequence A and B. j ,Δx j+1}, {Δx' j ,Δx' j+1} respectively represent the adjacent difference values ​​of the j-th opposite characteristic elements of sequences A and B. In this paper, I is set to 0.5 and J is usually set to -1.

[0093] In the improved set pair analysis method, the smaller the value obtained, the more obvious the smoothing effect between wind farms.

[0094] The steps of constructing the adjacency matrix using set pair analysis in this embodiment are as follows:

[0095] Step S11, using formula (2) to traverse and calculate the connection degree between different wind farms in the wind power cluster to obtain a connection degree matrix;

[0096] Step S12: Select the first m elements (m is the average number of connections of simultaneity and variance) in the connectivity matrix in ascending order, retain these m elements, and set the rest to zero to form a matrix M;

[0097] Step S13: Use the matrix M to construct the adjacency matrix A based on set pair analysis. s , the specific calculation process is as follows:

[0098]

[0099] Among them, μ ij is the connection degree between wind farms No. i and No. j. ∞ means there is no connection between the two wind farms. τ represents a minimum value, which is used to avoid the weight of the edge being 0. In this paper, it is set to 10. -5 .

[0100] In step S20, for the construction of the simultaneity rate adjacency matrix, the wind power simultaneity rate refers to the ratio of the highest active power of the wind farm during the statistical period to the total capacity of the grid-connected units during the period, which can be used to describe the evolution of the smoothing effect from the wind farm to the wind farm cluster. This paper uses the simultaneity rate to construct the adjacency matrix, but using the simultaneity rate of the entire sequence to construct the adjacency matrix will cause the time characteristics of the wind power output to be diluted or ignored. Therefore, this paper divides the wind power sequence into multiple subsequences and considers a method for constructing the adjacency matrix based on historical statistics. The specific method steps are as follows:

[0101] Step S21: traverse and calculate the daily synchronism rate of every two wind farms in the wind power cluster to form a matrix T k×i×j , the formula is as follows:

[0102]

[0103] in, represents the power sequence of the i-th wind farm on the k-th day, represents the power sequence of wind farm No. j on day k, Cap i Cap represents the operating capacity of wind farm No. i, j Represents the operating capacity of wind farm No. j.

[0104] Step S22: Matrix T k×i×j Discriminate each sub-matrix in the matrix to obtain the daily discriminant matrix, and sum up the daily discriminant matrices to obtain the final simultaneous rate discriminant matrix H i×j , the formula is as follows:

[0105]

[0106]

[0107] in, represents the discriminant matrix of the kth day, represents the judgment result of row i and column j on day k, h ij Represents the sum of the judgment results of the i-th row and j-th column of all days.

[0108] Step S23: Use the simultaneous rate discriminant matrix to calculate the adjacency matrix A of the directed weighted graph. t The specific formula is as follows:

[0109]

[0110] Among them, W ij It represents the weight of wind farm No. i pointing to wind farm No. j, and ∞ represents no connection (no edge).

[0111] In step S30, for the variance adjacency matrix, variance is used in probability theory and statistics to measure the degree of dispersion of a random variable or a set of data. To some extent, variance can also represent the degree of fluctuation of a sequence. Using variance to construct the adjacency matrix A v The process is the same as the simultaneous rate. It is only necessary to change the simultaneous rate matrix T in step S20 k×i×j The construction process is replaced by the variance matrix V k×i×j The construction process, the specific calculation formula is as follows:

[0112]

[0113]

[0114] in, represents the normalized sequence of the sum of wind farm No. i and wind farm No. j, Represents the mean of the sequence, and n is the number of elements in the sequence.

[0115] The final convergence effect adjacency matrix A a By A s , A t and A v The three adjacency matrices are calculated using the following formula:

[0116]

[0117] In step S40, the correlation coefficient adjacency matrix shows that wind farms located close together typically have similar outputs due to similar wind characteristics. Furthermore, wind farms in the same wind direction may also have a temporal lag in output. Therefore, this embodiment uses the classic Pearson correlation coefficient and Granger causality test to characterize the relationship between closely spaced wind farms. A corresponding adjacency matrix is ​​constructed simultaneously, and the two adjacency matrices are combined to produce the final temporal causal adjacency matrix.

[0118] The Pearson correlation coefficient is widely used to measure the degree of correlation between two variables, and its value is between -1 and 1. Use the Pearson correlation coefficient to construct the adjacency matrix A r The process is similar to that of the simultaneous rate, and formula (9) and formula (12) in step S20 are replaced by formulas (23) and (24).

[0119]

[0120] in, They represent the k-th day sequence of wind farm No. i and No. j respectively. They represent the mean values ​​of wind farm No. i and No. j on the kth day respectively.

[0121] In step S50, the Granger causality test, pioneered by 2003 Nobel Prize winner in Economics Clive WJ Granger, is applied to the causal adjacency matrix to analyze causal relationships between variables. Granger causality is specifically defined as follows: if the prediction of variable Y using past information about variables X and Y is better than the prediction based solely on past information about Y, that is, variable X helps explain future changes in variable Y, then variable X is considered to Granger cause variable Y. The specific steps are as follows:

[0122] Step S51: Regress the current y on all the lagged terms y and the remaining variables (if any), i.e., y on the lagged terms y of y. t-1 ,y t-2 ,...,y t-q The regression of x and other variables is done without including the lagged term x in this regression. This is a constrained regression. Then the constrained residual sum of squares RSS is obtained from this regression. R .

[0123] Step S52: Perform a regression with a lag term x, i.e., add the lag term x to the previous regression equation. This is an unconstrained regression, and the unconstrained residual sum of squares RSS is obtained from this regression. UR .

[0124] Step S53: The null hypothesis is H O :a1=a2=...=a q =0, that is, the lagged term x does not belong to this regression.

[0125] Step S54: To test this hypothesis, use the F test, namely: It follows an F-distribution with q and (nk) degrees of freedom. Here, n is the sample size, q is equal to the number of lags, i.e., the number of parameters to be estimated in the constrained regression equation, and k is the number of parameters to be estimated in the unconstrained regression equation.

[0126] Step S55: If the F value calculated at the selected significance level a exceeds the critical Fa value, the null hypothesis is rejected, and the lagged x term belongs to this regression, indicating that x is the cause of y.

[0127] Step S56: Similarly, in order to check whether y is the cause of x, the variables y and x can be replaced with each other, and steps S51-S55 can be repeated.

[0128] Use the Granger test to traverse and test whether the daily power series of different wind farms have Granger causality as shown in formula (25), and directly obtain the discriminant matrix of step S21 in step S20. Finally, use a similar method to step S20 to construct the Granger causality adjacency matrix A. g .

[0129]

[0130] The final temporal causal adjacency matrix A b By A r and A g Construction, calculation method and smoothing effect adjacency matrix A a same.

[0131] In step S60 , for the error characteristic adjacency matrix, there is also a smoothing effect on the errors between wind farms, and the manifestation of this smoothing effect is more obvious and direct. Figure 4The normalized errors between several groups of wind farms are shown in Figure 1. It is not difficult to find that there are a large number of positive and negative offset periods between the prediction errors of different wind farms. Although this positive and negative offset period does not benefit any individual wind farm, it can achieve prediction complementarity between wind farms and reduce the prediction error of wind power clusters. At the same time, this situation also illustrates the correlation between the two wind farms from the perspective of error. An error smoothing evaluation coefficient is proposed to quantify this relationship. The smaller the error smoothing evaluation coefficient, the stronger the prediction correlation between the two wind farms. The specific expression is as follows:

[0132]

[0133] in, They represent the k-th day forecast error series of wind farm i and wind farm j, respectively. If all forecast error series of two wind farms are used in the calculation, then they are the total error smoothing evaluation coefficients of the two wind farms.

[0134] Based on the error correlation between wind farms, a method for constructing an adjacency matrix is ​​proposed. In order to eliminate the interference of sequence length on error smoothing, the error smoothing evaluation coefficient is calculated every day and historical statistics are performed, similar to the correlation coefficient and Granger causality test. e The construction process and the simultaneous rate adjacency matrix A t Similarly, replace formula (9) in step S20 with formula (26).

[0135] Currently, graph neural networks have many applications in image and text tasks, and also in the field of wind power prediction. Because the spatial distribution of wind power clusters and the relationship between wind farms are very similar to the nodes and edges in graph neural networks, the application of graph neural networks in wind power cluster prediction is particularly suitable. Although there are many improved models in graph neural networks, the significance and prediction effect of graph convolutional neural networks as one of the classic basic models in graph neural networks cannot be ignored. The model function is defined as:

[0136]

[0137] Among them, H 0 =X,X∈R N*D , N is the number of nodes in the graph, D is the dimension of each node feature vector, H l ∈R N*d is the feature of the lth layer, H l+1 ∈R N*h is the updated feature of the l+1 layer, A is the adjacency matrix. D∈R N*N yes The degree matrix of W, d and h constitute the characteristic dimension of each node, representing the power or NWP time series data of a wind farm. l ∈R d*h is the weight parameter matrix of the lth layer. σ(·) is the nonlinear activation function.

[0138] Therefore, in step S70, a graph convolutional neural network based on the channel attention mechanism is proposed. The specific real-time steps are as follows:

[0139] Step S71: Using wind farms as nodes, design multiple graph convolution modules, inputting adjacency matrices of different attributes respectively to form a multi-channel model. Each channel is used to extract different characteristic rules in the wind power cluster to obtain multiple characteristics. The function definition is as follows:

[0140]

[0141] Among them, A a , A b and A e They represent the smoothing effect adjacency matrix, the causal adjacency matrix, and the error characteristic adjacency matrix respectively.

[0142] Step S72: Based on the multi-channel approach, an attention mechanism is introduced to learn the weight of each feature for prediction importance, and finally combine multiple features to obtain the final output. The function definition is as follows:

[0143]

[0144] Among them, Softmax(·) represents the normalization function, and concat(·) represents the concatenation function.

[0145] In one embodiment, the method for predicting the wind power cluster power day-ahead further includes steps S80 and S90 after step S70.

[0146] Step S80: simulation calculation.

[0147] Simulation input: measured power data of a wind power cluster containing multiple wind farms of sufficient length; day-ahead NWP data corresponding to the wind power cluster containing multiple wind farms; a data sampling interval of 15 minutes; and a day-ahead power forecast result of the wind power cluster is obtained according to steps S10 to S70.

[0148] Step S90: Error analysis.

[0149] The normalized root mean square error (NRMSE), normalized mean absolute error (NMAE) and coefficient of determination (R-squared, R 2 ) is used as the evaluation index of the prediction model performance, and the calculation formulas are:

[0150]

[0151] Where n is the length of the predicted power sequence, y i,pre and y i,ture are the predicted and actual values ​​of power, It represents the average value of actual power, and Cap is the installed capacity.

[0152] According to step S80, the simulation input is input, and the predicted power calculated by the model is compared with the measured power through the root mean square error formula (33), the mean absolute error formula (34) and the determination coefficient (35) in step 90 to perform error calculation.

[0153] To further demonstrate the feasibility and advancement of the present invention, this embodiment conducts experimental verification using three clusters: Wind farm cluster A, Wind farm cluster B, and Wind farm cluster C. Table 1 shows the wind farm cluster information.

[0154] Table 1 Wind power cluster information

[0155]

[0156] Based on the wind power cluster information shown in Table 1, the implementation steps of the method proposed in the present invention are as follows:

[0157] Step 1: If Figure 2 Demonstrates the methods and basis for distinguishing various relationships in set pair analysis. Figure 3 Shows how the strength of a feature can differ due to the difference between the values ​​of two series. Figure 3 (a) and Figure 3 (b) are opposite characteristics, but Figure 3 The intensities of the two opposite characteristics in (a) and (b) are quite different. Figure 3 (a)Δx a1 =Δx' a1 And Δx a2 =Δx' a2 , which is the most ideal case of opposite characteristics. Figure 3 (b)Δx b1 >Δx'b1 and Δxb2 >>Δx' b2 , the opposite characteristics are weaker. Although Δx d >>Δx c >Δx a ,but Figure 3 The opposite characteristics of (a), (c) and (d) are the same. Therefore, taking the historical power of each wind farm as the object, the connection degree between different wind farms within the wind power cluster is obtained through formulas (2)-(4). Then, using formulas (5) and (6), the adjacency matrix A of the improved set pair analysis method is calculated. s .

[0158] Step 2: Use formulas (7)-(16) to normalize the power data of each wind farm in the wind power cluster with the installed capacity as the denominator to obtain the simultaneous rate sequence, and then perform simultaneous rate judgment in units of days to obtain the simultaneous rate adjacency matrix A. t .

[0159] Step 3: With the simultaneous rate adjacency matrix A t The calculation process is similar to that of , replacing formulas (7)-(9) in formulas (7)-(16) with formulas (17)-(20) to obtain the variance adjacency matrix A v Then use formula (21) and formula (22) through A s , A t and A v Get the convergence effect adjacency matrix A a .

[0160] Step 4: Use Pearson correlation coefficient to construct the adjacency matrix A r The process is similar to the simultaneous rate adjacency matrix. Replace formula (9) and formula (12) in formula (7)-(16) with formula (23) and (24). Taking the power of each wind farm in the wind power cluster as input, the correlation coefficient adjacency matrix A can be obtained. r .

[0161] Step 5: Through the Granger causality test, the Granger test values ​​between the daily power values ​​of all wind farms in the wind power cluster are calculated to obtain a causal matrix similar to the formula (7) in the previous step S20. Then, the causal adjacency matrix A is calculated according to the process of formulas (10)-(16). g , where the discriminant formula (12) needs to be replaced with the causal discriminant formula (25). Finally, the correlation coefficient adjacency matrix A is used r and the causal adjacency matrix A g , we can get the temporal causal adjacency matrix A by taking the intersection of formula (21) and formula (22) b .

[0162] Step 6: Use the pooled effect adjacency matrix A obtained under the training set data a and the temporal causal adjacency matrix A b , input the test set NWP data for the first prediction, and obtain the prediction error value sequence. Then calculate the error characteristic adjacency matrix A through formulas (7)-(16) e , where formula (9) is replaced by the error evaluation characteristic coefficient formula (26).

[0163] Step 7: Combine the NWP features of each wind farm in the wind farm cluster under the test set and the three adjacency matrices A a , A b and A e Input the graph convolutional neural network based on the multi-channel attention mechanism, the structure is as follows Figure 4 , get the final power prediction value, and use formulas (33)-(35) to evaluate the prediction results.

[0164] Compared with the existing wind power cluster prediction strategy, the method of the present invention is more effective, and the prediction results are as follows: Figure 5 , the prediction error is shown in Table 2. The RMSE and MAE of the method proposed in this paper are the lowest among all clusters, R 2 The other cluster prediction methods performed differently within each wind farm cluster. However, looking at the average prediction errors across the three clusters, statistical upscaling performed the worst, with its RMSE being the only method to exceed 9%. This is because selecting a few representative wind farms within a wind farm cluster often fails to accurately represent the power trends of the entire cluster. However, due to the smaller number and concentrated distribution of wind farms in Cluster A, statistical upscaling performed better there. The holistic approach ignores many details within the cluster. While it reduces the prediction error compared to statistical upscaling, its improvement is very limited. The cumulative approach, on the other hand, fails to consider the inter-cluster correlations. Clustering, to some extent, addresses the shortcomings of the holistic and cumulative approaches, resulting in better prediction results. However, clustering does not consider inter-cluster correlations, and the classification method is not stable. Clustering significantly outperformed both the holistic and cumulative approaches in Cluster A, but did not offer a significant advantage in Clusters B and C. The graph convolutional model, on the other hand, takes the entire cluster as input and uses edges to represent the connections between wind farms. This approach truly balances the advantages of both the overall and local aspects, resulting in superior prediction results. In clusters A, B, and C, our method achieved average RMSE reductions of 1.55%, 1%, and 3.4%, respectively, compared to other prediction methods. Because cluster C has the largest number of wind farms and is the most widely distributed, the improvement achieved by our method is most pronounced there.

[0165] Table 2 Prediction errors of different clusters under each method

[0166]

[0167] Figure 5 (a) The high output samples in the second half of cluster A and Figure 5 In (c), at the peak output of cluster C, the prediction curves of other methods are mostly biased upwards, but the prediction curve of the method proposed in this paper can better track the changing trend of the actual power curve. Therefore, the prediction error of the method proposed in this paper has very little positive error (the error in which the predicted value is greater than the actual value). Figure 6 (a) Cluster A and Figure 6 In cluster C (c), the proposed method's prediction error accounts for a very small proportion of positive errors exceeding 5.5%, and is close to zero for positive errors exceeding 20%. Furthermore, the proposed method has a high proportion of low errors, demonstrating its superior power supply assurance capabilities compared to other methods. However, it is worth noting that the proposed method has a higher proportion of negative errors (errors where the predicted value is less than the actual value), but this does not affect its power supply assurance capabilities. The distribution of prediction errors for the proposed method in cluster B differs from that for clusters A and B. Figure 6 (b) shows that the proposed method has a certain proportion in each error range, but the proportion of small errors is higher, so the overall error is still the smallest. Figure 5 In (b), it is difficult to see a significant difference between the prediction curves of the proposed method and those of other methods. However, overall, the proposed method achieves good prediction results in all three clusters, with the most significant improvement in cluster C and a slight improvement in cluster B.

[0168] The above embodiments are only used to illustrate the present invention, and are not intended to limit the present invention. Ordinary technicians in the relevant technical field may make various changes and modifications without departing from the spirit and scope of the present invention. Therefore, all equivalent technical solutions also fall within the scope of the present invention. The scope of patent protection of the present invention should be defined by the claims.

Claims

1. A method for predicting wind power cluster power day-ahead, characterized in that: The method comprises: Based on the wind power sequence, construct the adjacency matrix A s , simultaneous rate adjacency matrix A t , variance adjacency matrix A v ; According to the adjacency matrix A s , simultaneous rate adjacency matrix A t and variance adjacency matrix A v , construct the convergence effect adjacency matrix A a ; Use Pearson correlation coefficient to construct the adjacency matrix A r , and construct the causal adjacency matrix A based on Granger causality check g ; Wherein, the adjacency matrix A r and the causal adjacency matrix A g Used to characterize the relationship between closely spaced wind farms; Based on the adjacency matrix A r and the Granger causality adjacency matrix A g , construct the temporal causal adjacency matrix A b ; Construct the error characteristic adjacency matrix A e ; Wherein, the error characteristic adjacency matrix A e Used to characterize the errors between wind farms; A graph convolutional neural network model based on the channel attention mechanism is constructed. In the graph convolutional neural network model, multiple graph convolution modules are designed to input adjacency matrices with different attributes to form multiple channels. Each channel is used to extract different characteristic laws in the wind power cluster and obtain multiple features. The function definition is as follows: Where, The l+1th layer features representing the convergence effect, The l+1th layer features representing temporal causality, Represents the l+1th layer feature of the error, f represents the graph convolution function, The l-th layer feature representing the convergence effect, The l-th layer features representing temporal causality, The lth layer feature representing the error; Based on multiple channels, the attention mechanism is introduced to learn the weight of each feature for prediction importance. Finally, multiple characteristics are combined to obtain the day-ahead wind power cluster power prediction result. The function is defined as follows: In the formula, Softmax(·) represents the normalization function, concat(·) represents the concatenation function, Output represents the day-ahead wind power cluster power forecast result, W a represents the weight of the convergence effect feature, W b Represents the weight of temporal causal features, W e represents the weight of the error feature, H represents the fusion feature, represents the learnable parameter of the l+1th layer convergence effect, represents the learnable parameters of the l+1th layer of temporal causality, Represents the learnable parameters of the error characteristics of the l+1th layer; The adjacency matrix A is constructed as follows s : The following formula is used to traverse and calculate the connection degree between different wind farms in the wind power cluster to obtain the connection degree matrix: Where, μ(A,B) is the connection matrix between wind power sequence A and wind power sequence B; {Δx i ,Δx i+1 }、{Δx′ i ,Δx′ i+1 } respectively represent the adjacent difference values ​​of the i-th identical characteristic element of wind power sequence A and wind power sequence B; {Δx j ,Δx j+1 }、{Δx′ j ,Δx′ j+1 } respectively represent the adjacent difference values ​​of the jth opposite characteristic elements of wind power sequence A and wind power sequence B; I is the difference coefficient; J is the opposition identification coefficient; N is the total number of features in the set; α i is the characteristic strength of the i-th identical set pair; β j is the characteristic strength of the jth opposite set pair; S′ is the total strength of the same set pair; F′ is the total strength of the uncertainty set pair; P′ is the total strength of the opposite set pair; In ascending order, select the first m elements in the connection matrix, where m is the average number of connections of the simultaneity rate and variance, retain m element values, and set the rest to zero to form a matrix M; Use matrix M to construct the adjacency matrix A based on set pair analysis s , the calculation process is as follows: Where μ ij is the connection degree between wind farm No. i and No. j; ∞ means there is no connection between the two wind farms; τ is a constant used to avoid the edge weight being 0; S ij Represents the edge weights established by set pair analysis.

2. The method according to claim 1, wherein The simultaneous rate adjacency matrix A is constructed as follows t : Traverse and calculate the daily synchronization rate of every two wind farms in the wind power cluster to form the matrix T k×i×j , the formula is as follows: Where, P i k represents the power sequence of the i-th wind farm on the k-th day, represents the power sequence of wind farm No. j on day k, Cap i Cap represents the operating capacity of wind farm No. i, j represents the operating capacity of wind farm No. j; represents the simultaneous rate matrix of the wind power cluster on the kth day, represents the simultaneity rate of the sum of the power of wind farms No. i and No. j on day k; For the matrix T k×i×j Discriminate each sub-matrix in the matrix to obtain the discriminant matrix of each day, and sum up the discriminant matrices of each day to obtain the simultaneous rate discriminant matrix H i×j , the formula is as follows: Where, represents the discriminant matrix of the kth day, represents the judgment result of row i and column j on day k, h ij Represents the sum of the judgment results of the i-th row and j-th column of all days; represents the simultaneity rate of wind farm No. j on day k; Using the simultaneous rate discriminant matrix to calculate the adjacency matrix A of the directed weighted graph t The specific formula for constructing is as follows: Where W ij represents the weight of wind farm No. i pointing to wind farm No. j; t ij Represents the edge weight established by the simultaneous rate.

3. The method according to claim 2, wherein The variance adjacency matrix A is constructed as follows v : Construct a variance matrix and construct a variance adjacency matrix A based on the variance matrix v ; Among them, the construction formula of the variance matrix is ​​as follows: Where V k×i×j represents the variance matrix, represents the normalized sequence of the sum of wind farm No. i and wind farm No. j, represents the sequence mean, n is the number of elements in the wind power sequence, w is the sequence number of the wind power sequence, represents the variance matrix of the wind power cluster on day k, It represents the variance of the sum of the power of wind farm No. i and No. j on day k.

4. The method according to claim 3, wherein According to the adjacency matrix A s , simultaneous rate adjacency matrix A t and variance adjacency matrix A v , the convergence effect adjacency matrix A is constructed by the following formula a : Where a ij represents the edge weight established by the convergence effect, s ij represents the edge weight constructed by the improved set pair analysis, t ij represents the edge weight constructed by the simultaneous rate, v ij Denotes the edge weights of the variance construction.

5. The method according to claim 1, wherein Use Pearson correlation coefficient to construct the adjacency matrix A r ,include: Traverse and calculate the correlation coefficient of every two wind farms in the wind power cluster every day to form a matrix P k×i×j , the formula is as follows: Where, represents the correlation coefficient matrix of the wind power cluster on day k; represents the correlation coefficient between the power of wind farm No. i and No. j on day k; denote the sequence of the kth day for wind farms i and j, respectively; represent the mean values ​​of wind farm No. i and No. j on the kth day respectively; For the matrix T k×i×j Discriminate each sub-matrix in the matrix to obtain the discriminant matrix of each day, and sum up the discriminant matrices of each day to obtain the simultaneous rate discriminant matrix H i×j , the formula is as follows: Where, represents the discriminant matrix of the kth day, represents the judgment result of row i and column j on day k, h ij Represents the sum of the judgment results of the i-th row and j-th column of all days; Using the simultaneous rate discriminant matrix to calculate the adjacency matrix A of the directed weighted graph r 's construction.

6. The method according to claim 1, wherein Constructing the causal adjacency matrix A based on Granger causality check g ,include: The Granger test is used to test whether the daily power series of different wind farms have Granger causality, thereby obtaining a discriminant matrix, and constructing a causal adjacency matrix A based on the discriminant matrix. g Among them, the calculation formula for using the Granger test to test whether the daily power series of different wind farms have Granger causality is as follows: Where, represents the Granger causality of wind farm No. i to wind farm No. j in the kth day; It represents the Granger test value of wind farm No. i on wind farm No. j in the kth day; Fa represents the Granger test threshold.

7. The method according to claim 2, wherein Construct the error characteristic adjacency matrix A e ,include: The error characteristic adjacency matrix A is obtained based on the error smoothing evaluation coefficient. e ; Wherein, the error smoothing evaluation coefficient is expressed as: Where, represents the normalized value of the sum of the errors of wind farms No. i and No. j on day k; They represent the k-th day prediction error series of wind farm No. i and wind farm No. j respectively.

8. The method according to any one of claims 1 to 7, characterized in that The method further comprises: Select normalized root mean square error NRMSE, normalized mean absolute error NMAE and coefficient of determination R 2 As the evaluation index of the prediction model performance, the calculation formulas are: Where n is the length of the predicted power sequence, y i,pre and y i,ture are the predicted and actual values ​​of power, It represents the average value of actual power, and Cap is the installed capacity.

Images