Wind power time sequence sample correction method and device considering time-space correlation, terminal equipment and storage medium

Abstract

The invention discloses a wind power time sequence sample correction method and device considering time-space correlation, terminal equipment and a storage medium, and belongs to the technical field of new energy, and the method comprises the steps: fitting a historical wind power data set, and generating a reference wind power time sequence sample; integrating all the reference wind power time sequence samples to form a total reference wind power time sequence sample; traversing each first time section of the to-be-corrected wind power time sequence sample, in the traversing process, taking a second time section with the minimum Euclidean distance from the current first time section as a target second time section, and correcting the current first time section based on second historical wind power data in the target second time section; and integrating all the corrected first time sections to obtain a historical wind power time sequence sample meeting spatial-temporal correlation. By implementing the method, the problem that a wind power time sequence sample in the prior art is difficult to accurately describe the space-time correlation of a large-scale offshore wind power cluster can be solved.

Description

Technical Field

[0001] The present invention relates to the field of new energy technologies, and in particular to a method, device, terminal equipment and storage medium for correcting wind power time series samples taking into account temporal and spatial correlation. Background Art

[0002] Wind power data is an essential component of steady-state simulations and calculations for power systems that include wind power. However, calculations related to power system planning often require wind power data spanning decades or even centuries, and this data is not readily available in open resources. Therefore, the mainstream approach to constructing wind power cluster scenarios is to build probability distribution information based on historical data samples, and commonly employ probabilistic models such as the Weibull distribution and the Beta distribution to describe the stochastic characteristics of wind and photovoltaic power output. Some studies have also employed the Markov Chain Monte Carlo method (MCMC) to introduce Markov chains from random processes into Monte Carlo simulations.

[0003] However, it should be pointed out that probability models such as the Webull distribution and the Beta distribution have certain limitations. They do not fully consider the state transition and persistence characteristics of time series. In addition, although the Markov Chain Monte Carlo method can better describe the temporal characteristics of the output of a single wind farm, in the actual scenario of large-scale offshore wind power cluster output, multiple wind farms in the region each have unique probability characteristics and temporal characteristics, making it difficult for the Markov Chain Monte Carlo method to accurately describe the spatiotemporal correlation of large-scale offshore wind power clusters. The wind power time series samples generated based on this cannot meet the actual needs of large-scale offshore wind power cluster planning, scheduling and operation. Summary of the Invention

[0004] The present invention provides a method, device, terminal device and storage medium for correcting wind power time series samples taking into account temporal and spatial correlation. The method can solve the problem in the prior art that wind power time series samples are difficult to accurately describe the temporal and spatial correlation of large-scale offshore wind power clusters.

[0005] An embodiment of the present invention provides a method for correcting wind power time series samples taking into account temporal and spatial correlation, comprising:

[0006] Acquire correction parameters, time series samples of wind power to be corrected, and historical wind power data sets of each wind farm; wherein the time series samples of wind power to be corrected include a plurality of first time sections;

[0007] For each wind farm, fitting the historical wind power data set of the wind farm to obtain a Gaussian mixture model for characterizing the distribution characteristics of the time-series output curve, and calculating a reference wind power time-series sample based on an initial correlation coefficient matrix, the correction parameters, and the Gaussian mixture model; wherein the initial correlation coefficient matrix is ​​calculated using the historical wind power data set;

[0008] Integrate the reference wind power time series samples of each wind farm to obtain the total reference wind power time series samples;

[0009] Traversing each first time section, during each traversal, calculating the Euclidean distance between the first historical wind power data corresponding to the current first time section and the second historical wind power data corresponding to each second time section in the current total reference wind power time series sample, selecting the second time section with the smallest Euclidean distance as the target second time section, correcting the current first time section according to the second historical wind power data corresponding to the target second time section to obtain a corrected first time section, and removing the target second time section from the current total reference wind power time series sample;

[0010] By integrating all the corrected first time sections, we can obtain the historical wind power time series samples that meet the spatiotemporal correlation.

[0011] Furthermore, the method for obtaining the historical wind power data set includes:

[0012] Obtaining historical wind power data of the wind farm within a preset first time period;

[0013] Normalizing the historical wind power data to obtain normalized historical wind power data;

[0014] The historical wind power data is divided using the time characteristics of the normalized historical wind power data as a classification basis and using the length of a preset second time period as a separation scale to obtain a plurality of historical wind power data sets to be smoothed; wherein the length of the preset second time period is less than the length of the preset first time period; and the time periods covered by the historical wind power data sets to be smoothed do not overlap with each other;

[0015] For each of the historical wind power data sets to be smoothed, performing data smoothing processing on the historical wind power data set to be smoothed to obtain a historical wind power data set after data smoothing processing;

[0016] The historical wind power data set after the data smoothing process is used as the historical wind power data set.

[0017] Furthermore, performing data smoothing on the historical wind power data set to be smoothed to obtain the historical wind power data set after data smoothing includes:

[0018] Dividing the historical wind power data set to be smoothed into a plurality of first equidistant intervals;

[0019] Counting the total number of historical wind power data in each first equidistant interval to obtain a first total number;

[0020] Identifying invalid data in each first equidistant interval, removing the invalid data from the corresponding first equidistant interval, and using the first equidistant interval after removing the invalid data as the second equidistant interval;

[0021] Counting the total number of invalid data to obtain a second total number;

[0022] According to the first total number, the second total number, and a preset optimization model, a number of samples that can be filled in a preset third equidistant interval is obtained; wherein the second equidistant interval and the third equidistant interval do not overlap with each other;

[0023] Taking the number of fillable samples as the filling upper limit, randomly generating wind power data samples and filling them into the third equidistant interval to obtain the third equidistant interval after filling;

[0024] Based on the third equidistant interval after filling and the second equidistant interval, a historical wind power data set after data smoothing is reconstructed; wherein the second equidistant interval and the third equidistant interval are located on the same reference coordinate axis.

[0025] Furthermore, the calculation of the reference wind power time series sample based on the initial correlation coefficient matrix, the correction parameter and the Gaussian mixture model includes:

[0026] Generating an initial reference wind power time series sample according to a probability density function corresponding to the Gaussian mixture model;

[0027] Normalizing the initial wind power time series samples to obtain normalized reference wind power time series samples;

[0028] The normalized reference wind power time series samples are corrected according to the initial correlation coefficient matrix, the correction parameters and the Gaussian mixture model to obtain final reference wind power time series samples.

[0029] Furthermore, the correction parameters include the maximum number of iterations, convergence accuracy and change step size;

[0030] The step of correcting the normalized reference wind power time series samples according to the initial correlation coefficient matrix, the correction parameter, and the Gaussian mixture model to obtain a final reference wind power time series sample includes:

[0031] Repeating the iterative calculation operation according to the correction parameter, the initial correlation coefficient matrix and the Gaussian mixture model until the final reference wind power time series sample is obtained;

[0032] The iterative calculation operation includes:

[0033] Update the current number of iterations;

[0034] Decomposing the current correlation coefficient matrix to obtain a lower triangular matrix; wherein, when performing the iterative calculation operation for the first time, the current correlation coefficient matrix is ​​set as the initial correlation coefficient matrix;

[0035] According to the preset sample matrix obeying the standard normal distribution and the lower triangular matrix, a standard normal distribution sample obeying the correlation is calculated;

[0036] Mapping the standard normal distribution samples to the original sample space that conforms to the multivariate normal distribution in the Gaussian mixture model to obtain candidate wind power time series samples;

[0037] Calculating an approximate correlation coefficient matrix based on the candidate wind power time series samples;

[0038] Calculating the difference between the approximate correlation coefficient matrix and the initial correlation coefficient matrix;

[0039] Determine whether the difference is not less than the convergence accuracy, and determine whether the current number of iterations is less than the maximum number of iterations,

[0040] If the judgment results are all yes, the current correlation coefficient matrix is ​​corrected according to the difference and the change step size to obtain a corrected correlation coefficient matrix, and the corrected correlation coefficient matrix is ​​used as the current correlation coefficient matrix for the next iterative calculation operation;

[0041] Otherwise, the candidate wind power time series sample is used as the final reference wind power time series sample.

[0042] Furthermore, mapping the standard normal distribution samples to the original sample space conforming to the multivariate normal distribution in the Gaussian mixture model to obtain candidate wind power time series samples includes:

[0043] Substituting each element in the standard normal distribution sample into the preset standard normal distribution cumulative distribution function, and calculating the probability value corresponding to each element;

[0044] For each element in the standard normal distribution sample, according to the probability value of the element, an approximate value of the probability value is found in a preset table as the element of the alternative wind power time series sample; wherein, the preset table includes a number of equidistant points selected from the historical wind power data set and the cumulative distribution function value corresponding to each equidistant point; the cumulative distribution function value corresponding to each equidistant point is calculated by substituting each equidistant point into the Gaussian mixture model respectively.

[0045] Furthermore, the correcting the current first time section according to the second historical wind power data corresponding to the target second time section includes:

[0046] The first historical wind power data corresponding to the current first time section is replaced by the second historical wind power data corresponding to the target second time section.

[0047] An embodiment of the present invention further provides a wind power time series sample correction device taking into account temporal and spatial correlation, comprising: a data acquisition module, a reference wind power time series sample determination module, a total reference wind power time series sample determination module, a correction module, and a historical wind power time series sample generation module;

[0048] The data acquisition module is used to acquire correction parameters, time series samples of wind power to be corrected, and historical wind power data sets of each wind farm; wherein the time series samples of wind power to be corrected include a plurality of first time sections;

[0049] The reference wind power time series sample determination module is configured to fit the historical wind power data set of each wind farm to obtain a Gaussian mixture model for characterizing the distribution characteristics of the time series output curve, and calculate the reference wind power time series sample based on the initial correlation coefficient matrix, the correction parameters, and the Gaussian mixture model; wherein the initial correlation coefficient matrix is ​​calculated using the historical wind power data set;

[0050] The total reference wind power time series sample determination module is used to integrate the reference wind power time series samples of each wind farm to obtain the total reference wind power time series sample;

[0051] The correction module is used to traverse each first time section, and during each traversal, calculate the Euclidean distance between the first historical wind power data corresponding to the current first time section and the second historical wind power data corresponding to each second time section in the current total reference wind power time series sample, select the second time section with the smallest Euclidean distance as the target second time section, correct the current first time section according to the second historical wind power data corresponding to the target second time section, obtain a corrected first time section, and remove the target second time section from the current total reference wind power time series sample;

[0052] The historical wind power time series sample generation module is used to integrate all the corrected first time sections to obtain historical wind power time series samples that meet the time-space correlation.

[0053] The present application also provides a terminal device, including:

[0054] one or more processors;

[0055] a memory, coupled to the processor, for storing one or more programs;

[0056] When the one or more programs are executed by the one or more processors, the one or more processors implement the wind power time series sample correction method taking into account the temporal and spatial correlation as described in the above embodiment of the invention.

[0057] The present application also provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the method for correcting wind power time series samples taking into account the temporal and spatial correlation as described in the above-mentioned embodiment of the invention is implemented.

[0058] The following beneficial effects are achieved by implementing the present invention:

[0059] The present invention provides a wind power time series sample correction method, device, terminal device and storage medium taking into account the temporal and spatial correlation. The method first fits the historical wind power data set of the wind farm to obtain a Gaussian mixture model for characterizing the distribution characteristics of the time series output curve. The Gaussian mixture model can accurately describe the probability distribution law of the wind farm output in the time dimension, laying the foundation for the subsequent consideration of the time correlation (autocorrelation) of the wind farm itself; then, through the initial correlation coefficient matrix, correction parameters and Gaussian mixture model, a reference wind power time series sample is calculated so that the reference The wind power time series samples can accurately reflect the temporal correlation (autocorrelation); then, the reference wind power time series samples of each wind farm are integrated to obtain the total reference wind power time series samples. In this process, the relationship between the reference wind power time series samples of different wind farms is fully considered, and the spatial correlation (cross-correlation) between each wind farm is taken into account; finally, based on the total reference wind power time series samples, the first time sections of the wind power time series samples to be corrected are corrected, and all the corrected first time sections are integrated to obtain the historical wind power time series samples that meet the temporal and spatial correlation;

[0060] Therefore, the historical wind power time series samples generated after correction take into account both the temporal correlation of a single wind farm itself and the spatial correlation between wind farms, effectively solving the problem in existing technologies that wind power output data do not consider temporal and spatial correlations, and can provide more accurate data support for large-scale offshore wind power cluster output modeling. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] In order to more clearly illustrate the technical solution of the present application, the following is a brief introduction to the drawings required for use in the implementation. Obviously, the drawings described below are only some implementation methods of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0062] Figure 1 This is a flowchart of a method for correcting wind power time series samples taking into account temporal and spatial correlations provided by a certain embodiment of the present application;

[0063] Figure 2 This is another flowchart of a method for correcting wind power time series samples taking into account temporal and spatial correlations provided by a certain embodiment of the present application;

[0064] Figure 3 This is a schematic diagram of historical wind power data of four offshore wind farms in a coastal province in 2020, provided in an embodiment of the present application;

[0065] Figure 4This is a schematic diagram of historical wind power data of four offshore wind farms in a coastal province in 2021, provided in an embodiment of the present application;

[0066] Figure 5 This is a schematic diagram of the effect of the Gaussian mixture model obtained by fitting provided in an embodiment of the present application;

[0067] Figure 6 It is a probability density comparison diagram between the sample of offshore wind farm 1 and the total reference wind power time series sample in the historical wind power time series sample provided by a certain embodiment of the present application;

[0068] Figure 7 It is a probability density comparison diagram between the sample of offshore wind farm 2 and the total reference wind power time series sample in the historical wind power time series sample provided by a certain embodiment of the present application;

[0069] Figure 8 It is a probability density comparison diagram between the sample of offshore wind farm 3 in the historical wind power time series sample provided by a certain embodiment of the present application and the total reference wind power time series sample;

[0070] Figure 9 It is a probability density comparison diagram between the sample of offshore wind farm 4 and the total reference wind power time series sample in the historical wind power time series sample provided by a certain embodiment of the present application;

[0071] Figure 10 This is a schematic diagram of the spatial correlation satisfied between offshore wind farms in a historical wind power time series sample provided by an embodiment of the present application;

[0072] Figure 11 is a schematic diagram of the spatial correlation satisfied between the offshore wind farms in the total reference wind power time series sample provided by an embodiment of the present application;

[0073] Figure 12 is a load power curve of a normalized annual historical wind power time series sample provided in a certain embodiment of the present application;

[0074] Figure 13 is a wind power curve of offshore wind farm 1 in the normalized annual historical wind power time series sample provided in a certain embodiment of the present application;

[0075] Figure 14 is a wind power curve of offshore wind farm 2 in the normalized annual historical wind power time series sample provided in a certain embodiment of the present application;

[0076] Figure 15 is a wind power curve of offshore wind farm 3 in the normalized annual historical wind power time series sample provided in a certain embodiment of the present application;

[0077] Figure 16 is a wind power curve of the offshore wind farm 4 in the normalized annual historical wind power time series sample provided in a certain embodiment of the present application;

[0078] Figure 17 This is a structural diagram of a wind power time series sample correction device taking into account temporal and spatial correlation provided by a certain embodiment of the present application;

[0079] Figure 18 This is a schematic diagram of the structure of a terminal device provided in a certain embodiment of the present application. DETAILED DESCRIPTION

[0080] To make the objectives, technical solutions, and advantages of this application more clear, the technical solutions in this application will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of this application. Obviously, the described embodiments are only part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making any creative efforts are within the scope of protection of this application.

[0081] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which this application belongs; the terms used herein are only for the purpose of describing specific embodiments and are not intended to limit this application; the terms "including" and "having" and any variations thereof in the specification and claims of this application and the above-mentioned figure descriptions are intended to cover non-exclusive inclusions.

[0082] In the description of the embodiments of this application, the technical terms "first" and "second" are used only to distinguish different objects and should not be understood to indicate or imply relative importance or implicitly specify the quantity, specific order, or primary and secondary relationship of the indicated technical features. In the description of the embodiments of this application, the meaning of "plurality" is more than two, unless otherwise clearly and specifically defined.

[0083] References herein to "embodiments" mean that a particular feature, structure, or characteristic described in connection with the embodiments may be included in at least one embodiment of the present application. The appearance of this phrase in various places in the specification does not necessarily refer to the same embodiment, nor does it constitute an independent or alternative embodiment that is mutually exclusive of other embodiments. It is understood, both explicitly and implicitly, by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0084] In the description of the embodiments of this application, the term "and / or" is simply a description of the association relationship between associated objects, indicating that three relationships can exist. For example, A and / or B can represent the following three situations: A exists alone, A and B exist simultaneously, and B exists alone. In addition, the character " / " in this document generally indicates that the associated objects are in an "or" relationship.

[0085] In the description of the embodiments of the present application, the term "multiple" refers to more than two (including two). Similarly, "multiple groups" refers to more than two groups (including two groups), and "multiple pieces" refers to more than two pieces (including two pieces).

[0086] In the description of the embodiments of the present application, unless otherwise expressly specified or limited, technical terms such as "installed," "connected," "connected," and "fixed" should be understood in a broad sense. For example, they can refer to fixed connections, detachable connections, or integration; mechanical connections or electrical connections; direct connections or indirect connections through an intermediate medium; internal connections between two components or interactions between two components. Those skilled in the art can understand the specific meanings of the above terms in the embodiments of the present application based on specific circumstances.

[0087] See Figure 1 To solve the problem in the prior art that wind power time series samples are difficult to accurately describe the spatiotemporal correlation of large-scale offshore wind power clusters, an embodiment of the present invention provides a wind power time series sample correction method that takes into account the spatiotemporal correlation, including:

[0088] S1. Obtain correction parameters, time series samples of wind power to be corrected, and historical wind power data sets of each wind farm; wherein the time series samples of wind power to be corrected include a plurality of first time sections;

[0089] Specifically, it is assumed that there are currently A wind farms; the wind power time series samples to be corrected have been generated in advance by existing methods such as Markov Chain Monte Carlo (MCMC), covering the wind power data records of A wind farms in a certain year;

[0090] It should be noted that, due to the limitations of the generation method, there are two possible correlation states for the wind power time series samples to be corrected: first, it only presents the correlation of the output of a single wind farm itself in the time dimension (i.e., time correlation), but lacks the correlation between the wind farms in the spatial dimension (i.e., spatial correlation); second, it is completely unable to reflect any correlation in the time and space dimensions; here, we will not limit the specific generation method of the wind power time series samples to be corrected, but only emphasize the correlation state presented after its generation.

[0091] In a preferred embodiment, the method for acquiring the historical wind power data set includes:

[0092] Obtaining historical wind power data of the wind farm within a preset first time period;

[0093] Normalizing the historical wind power data to obtain normalized historical wind power data;

[0094] The historical wind power data is divided using the time characteristics of the normalized historical wind power data as a classification basis and using the length of a preset second time period as a separation scale to obtain a plurality of historical wind power data sets to be smoothed; wherein the length of the preset second time period is less than the length of the preset first time period; and the time periods covered by the historical wind power data sets to be smoothed do not overlap with each other;

[0095] For each of the historical wind power data sets to be smoothed, performing data smoothing processing on the historical wind power data set to be smoothed to obtain a historical wind power data set after data smoothing processing;

[0096] The historical wind power data set after the data smoothing process is used as the historical wind power data set;

[0097] Illustratively, historical wind power data of a wind farm A in a certain year is obtained. In this embodiment, the first period is preset to be annual, which extends from the beginning of the year to the end of the year, covering a full year.

[0098] Specifically, taking the historical wind power data of the i-th wind farm in a certain year 1 as an example, assuming that the statistical value of the historical wind power data of the i-th wind farm in a certain year is P′ (i) , in obtaining P′ (i) After that, it is necessary to remove the obviously abnormal data (but not to remove the 0 element), and then correct and fill the missing values ​​in the data; then calculate P′ (i) The maximum value in and minimum value For P′ (i) Perform normalization processing to obtain the normalized data P (i) , where the specific formula for normalization is as follows:

[0099]

[0100] Illustratively, in this embodiment, the preset second period is set to a month;

[0101] Specifically, the time characteristics (such as timestamps, etc.) of the normalized historical wind power data are used as the classification basis, and the time length corresponding to each month in year 1 is used as the separation scale to divide the historical wind power data. Since there are 12 months in a year, the historical wind power data for the whole year will be classified according to the time length of each month in the order of the months, and 12 historical wind power data sets to be smoothed are obtained.

[0102] Indicatively, since each of the obtained historical wind power data sets to be smoothed may contain multiple zero elements, these zero elements will cause abnormal fluctuations in the data distribution, interfering with the Gaussian mixture model's capture of the true distribution characteristics of the data, making it difficult for the Gaussian mixture model to accurately describe the regularity of the wind power data in the time series. Therefore, it is necessary to perform data smoothing processing on each of the historical wind power data sets to be smoothed:

[0103] In a preferred embodiment, performing data smoothing on the historical wind power data set to be smoothed to obtain the historical wind power data set after data smoothing includes:

[0104] Dividing the historical wind power data set to be smoothed into a plurality of first equidistant intervals;

[0105] Counting the total number of historical wind power data in each first equidistant interval to obtain a first total number;

[0106] Identifying invalid data in each first equidistant interval, removing the invalid data from the corresponding first equidistant interval, and using the first equidistant interval after removing the invalid data as the second equidistant interval;

[0107] Counting the total number of invalid data to obtain a second total number;

[0108] According to the first total number, the second total number, and a preset optimization model, the number of samples that can be filled in a preset third equidistant interval is solved; wherein the second equidistant interval and the third equidistant interval do not overlap with each other;

[0109] Taking the number of fillable samples as the filling upper limit, randomly generating wind power data samples and filling them into the third equidistant interval to obtain the third equidistant interval after filling;

[0110] reconstructing a historical wind power data set after data smoothing based on the third equidistant interval and the second equidistant interval after the filling is completed; wherein the second equidistant interval and the third equidistant interval are located on the same reference coordinate axis;

[0111] For example, the smoothed historical wind power data set P of the i-th wind farm in the j-th month is used. (i,j)For example, the historical wind power data set P to be smoothed (i,j) Divide into N first equidistant intervals, and the interval between each first equidistant interval is 1 / N;

[0112] Specifically, the first total number of historical wind power data falling in the first first equidistant interval [0, 1 / N) is n1, the first total number of historical wind power data falling in the second first equidistant interval [1 / N, 2 / N) is n2, ..., the first total number of historical wind power data falling in the Nth first equidistant interval [1-(1 / N), 1] is n N .

[0113] Indicatively, invalid data refers to zero elements. To better balance data distribution and reduce the interference of zero elements on the overall data set, four third intervals are pre-divided. After identifying the zero elements, they are dispersed to each third interval while ensuring uniform distribution, thereby weakening the concentration effect of zero elements and improving the overall stability and representativeness of the data.

[0114] Specifically, after identifying the 0 elements in each first equidistant interval, the 0 elements are removed from each first equidistant interval, and the first equidistant interval after removing the invalid data is used as the second equidistant interval, and the total number of invalid data is counted to obtain a second total number N0;

[0115] It should be noted that the constructed third equidistant interval and the second equidistant interval are located on the same reference coordinate axis. The first third equidistant interval is located in the interval [-1 / N, 0), the second third equidistant interval is located in the interval [-2 / N, -1 / N), the third third equidistant interval is located in the interval [-3 / N, -2 / N), and the fourth third equidistant interval is located in the interval [-4 / N, -3 / N);

[0116] Specifically, the first total number n1 of the first first equidistant interval is subtracted from the second total number N0 to obtain the difference n′1=n1-N0. Assuming that the number of fillable samples falling into the first third equidistant interval is N -1 , the number of samples that can be filled in the second and third equal intervals is N -2 , the number of samples that can be filled in the third third equal interval is N -3 , the number of samples that can be filled in the fourth third equal interval is N -4 In order to obtain the specific value of the number of samples that can be filled in each third equidistant interval, the following optimization model is constructed:

[0117]

[0118] Wherein, Δh represents a preset proportional parameter; θ represents a preset temporary variable, and θ=N -1Δh;

[0119] And set constraints in the optimization model to limit N -4 、N -3 、N -2 and N -1 The following constraints must be met:

[0120] N -3 =N -4 Δh;

[0121] N -2 =N -3 Δh;

[0122] N -1 =N -2 Δh;

[0123] N -4 +N -3 +N -2 +N -1 =N0;

[0124] n′1≥N -1 Δh;

[0125] Therefore, according to the optimization model and the internal constraints, the preset simulation software is called to solve the problem, thereby solving N -4 、N -3 、N -2 and N -1 The specific value of .

[0126] In an illustrative manner, wind power data samples are randomly generated and filled into the third equidistant interval to obtain the third equidistant interval after filling;

[0127] Specifically, the wind power data samples filled in the first third equidistant interval are:

[0128]

[0129] The wind power data samples filled in the second and third equidistant intervals are:

[0130]

[0131] The wind power data sample filled in the third equidistant interval is:

[0132]

[0133] The wind power data samples filled in the fourth third equidistant interval are:

[0134]

[0135] It should be noted that the randomly generated wind power data samples are data samples with random numbers between 0 and 1, that is, rand∈[0,1];

[0136] Thus, the randomly generated wind power data samples are filled into the corresponding third equidistant interval to obtain the third equidistant interval after filling. Based on the third equidistant interval after filling and the second equidistant interval, the historical wind power data set of the i-th wind farm in the j-th month after data smoothing is reconstructed.

[0137] S2. For each wind farm, fitting the historical wind power data set of the wind farm to obtain a Gaussian mixture model for characterizing the distribution characteristics of the time-series output curve, and calculating a reference wind power time-series sample based on an initial correlation coefficient matrix, the correction parameters, and the Gaussian mixture model; wherein the initial correlation coefficient matrix is ​​calculated using the historical wind power data set;

[0138] Schematically, the historical wind power data sets of each wind farm are fitted to obtain the Gaussian mixture model corresponding to each wind farm for characterizing the distribution characteristics of the time series output curve;

[0139] Specifically, the historical wind power data set of the i-th wind farm in the j-th month is For example, the historical wind power data set is processed by the fitgmdistha function in MATLAB. Perform fitting to obtain the historical wind power data set The corresponding Gaussian mixture model, in this process, the solution algorithm adopts the EM iteration method, where the expressions of the probability density function p(x) and cumulative distribution function F(x) corresponding to the Gaussian mixture model are as follows:

[0140]

[0141]

[0142] Where K represents the number of Gaussian distributions (i.e., the number of clusters), π k represents the mixture weight of the k-th Gaussian distribution, μ k represents the mean vector of the kth Gaussian distribution, σ k represents the standard deviation of the k-th Gaussian distribution, N(x|μ k ,σ k ) represents the probability density function of the kth Gaussian distribution at x, Φ k (x) represents the cumulative distribution function of the k-th Gaussian distribution at x.

[0143] In a preferred embodiment, the calculation of the reference wind power time series sample based on the initial correlation coefficient matrix, the correction parameter and the Gaussian mixture model includes:

[0144] Generating an initial reference wind power time series sample according to a probability density function corresponding to the Gaussian mixture model;

[0145] Normalizing the initial wind power time series samples to obtain normalized reference wind power time series samples;

[0146] Correcting the normalized reference wind power time series samples according to the initial correlation coefficient matrix, the correction parameters, and the Gaussian mixture model to obtain final reference wind power time series samples;

[0147] Schematically, based on the probability density function p(x) corresponding to the Gaussian mixture model, an initial reference wind power time series sample X that obeys the correlation is generated. (i,j) , and the initial reference wind power time series sample X (i,j) Normalize and obtain the normalized reference wind power time series sample

[0148] Specifically, the normalization operation is: determining the initial reference wind power time series sample X (i,j) For elements greater than 1 or less than 0, set the elements greater than 1 to 1 and the elements less than 0 to 0.

[0149] Schematic, normalized reference wind power time series sample The correlation will inevitably be changed, so it is necessary to normalize the reference wind power time series samples after normalization. Make corrections;

[0150] Specifically, in a preferred embodiment, the correction parameters include the maximum number of iterations, convergence accuracy and change step size;

[0151] The step of correcting the normalized reference wind power time series samples according to the initial correlation coefficient matrix, the correction parameter, and the Gaussian mixture model to obtain a final reference wind power time series sample includes:

[0152] Repeating the iterative calculation operation according to the correction parameter, the initial correlation coefficient matrix and the Gaussian mixture model until the final reference wind power time series sample is obtained;

[0153] The iterative calculation operation includes:

[0154] Update the current number of iterations;

[0155] Decomposing the current correlation coefficient matrix to obtain a lower triangular matrix; wherein, when performing the iterative calculation operation for the first time, the current correlation coefficient matrix is ​​set as the initial correlation coefficient matrix;

[0156] According to the preset sample matrix obeying the standard normal distribution and the lower triangular matrix, a standard normal distribution sample obeying the correlation is calculated;

[0157] Mapping the standard normal distribution samples to the original sample space that conforms to the multivariate normal distribution in the Gaussian mixture model to obtain candidate wind power time series samples;

[0158] Calculating an approximate correlation coefficient matrix based on the candidate wind power time series samples;

[0159] Calculating the difference between the approximate correlation coefficient matrix and the initial correlation coefficient matrix;

[0160] Determine whether the difference is not less than the convergence accuracy, and determine whether the current number of iterations is less than the maximum number of iterations,

[0161] If the judgment results are all yes, the current correlation coefficient matrix is ​​corrected according to the difference and the change step size to obtain a corrected correlation coefficient matrix, and the corrected correlation coefficient matrix is ​​used as the current correlation coefficient matrix for the next iterative calculation operation;

[0162] Otherwise, the candidate wind power time series sample is used as the final reference wind power time series sample;

[0163] Specifically, the following are the steps of the iterative calculation operation:

[0164] Step S201, define correction parameters and initialize: define the current correlation coefficient matrix R on the standard normal distribution domain z The initial value is Initialize the number of iterations p = 0, convergence accuracy ε = 0.01, and the maximum number of iterations p max =100, the change step η = 0.1 ~ 0.5; where, is the initial correlation coefficient matrix;

[0165] Step S202: Update the number of iterations, p=p+1;

[0166] Step S203: Calculate the standard normal distribution sample Z' new :Use Cholesky decomposition method to decompose the current correlation coefficient matrix The lower triangular matrix L is Based on the lower triangular matrix L, the standard normal distribution sample Z′ that obeys the correlation is calculated new =L·Z new , where Z new Represents a preset sample matrix that obeys the standard normal distribution. The number of rows in the sample matrix is ​​the total number of wind farms, and the number of columns is the number of hours contained in the jth month. It is generated using the randn function in MATLAB. Its matrix dimension is the same as the reference wind power time series sample to be generated. Stay consistent;

[0167] Step S204: Calculate candidate wind power time series samples The standard normal distribution sample Z' new Mapped to the original sample space that conforms to the multivariate normal distribution in the Gaussian mixture model, the alternative wind power time series samples are obtained Among them, G represents the cumulative distribution function of the preset standard normal distribution, F -1 Represents the inverse function of the cumulative distribution function of the Gaussian mixture model;

[0168] Step S205: Calculate the approximate correlation coefficient matrix Screen out alternative wind power time series samples Set the 0 element in the , and calculate the alternative wind power time series sample The corresponding approximate correlation coefficient matrix (Pearson correlation coefficient matrix);

[0169] Step S206: Determine the convergence accuracy and number of iterations: Calculate the approximate correlation coefficient matrix With the initial correlation coefficient matrix The difference When the conditions are met at the same time and p<p max When , go to step S207, otherwise, go to step S208;

[0170] Step S207: Modify the current correlation coefficient matrix pass Modify the current correlation coefficient matrix Get the current correlation coefficient matrix for the next iteration And go to step S202;

[0171] Step S208: Output reference wind power time series samples The alternative wind power time series sample As the final reference wind power time series sample, it jumps out of the iteration loop.

[0172] In a preferred embodiment, mapping the standard normal distribution samples to the original sample space conforming to the multivariate normal distribution in the Gaussian mixture model to obtain candidate wind power time series samples includes:

[0173] Substituting each element in the standard normal distribution sample into the preset standard normal distribution cumulative distribution function, and calculating the probability value corresponding to each element;

[0174] For each element in the standard normal distribution sample, according to the probability value of the element, find an approximate value of the probability value in a preset table, and use it as the element of the candidate wind power time series sample; wherein the preset table includes a number of equally spaced points selected from the historical wind power data set and the cumulative distribution function value corresponding to each equally spaced point; the cumulative distribution function value corresponding to each equally spaced point is calculated by substituting each equally spaced point into the Gaussian mixture model;

[0175] Specifically, in order to efficiently obtain F -1 First, build a preset table: From the historical wind power data set Taking enough point elements (e.g., 1000 point elements) in the second and third equally spaced intervals, and then obtaining the cumulative distribution function value corresponding to each point element in the Gaussian mixture model;

[0176] Then, based on the selected point elements and the calculated cumulative distribution function values ​​corresponding to the point elements in the Gaussian mixture model, the preset table is constructed, wherein each point element in the preset table has a one-to-one correspondence with the cumulative distribution function value calculated therefrom;

[0177] After constructing the preset table, the standard normal distribution sample Z' new Substitute each element in the cumulative distribution function of the preset standard normal distribution to calculate the probability value G(Z′) corresponding to each element new ), then according to G(Z′ new ) value, select the value that matches G(Z′) in the preset table by looking up the table. new ) is the nearest value, and as an approximation, we get F-1(G(Z′ new )),

[0178] The cumulative distribution function of the preset standard normal distribution is expressed as follows:

[0179]

[0180] Among them, z′ new,ij Represents Z′ new The element at row i and column j in .

[0181] Therefore, based on the above steps, the candidate wind power time series sample is obtained. elements of .

[0182] S3. Integrate the reference wind power time series samples of each wind farm to obtain a total reference wind power time series sample;

[0183] Specifically, the reference wind power time series samples of all wind farms in month j are integrated to obtain the total reference wind power time series samples of month j:

[0184] S4. Traverse each first time section, and during each traversal, calculate the Euclidean distance between the first historical wind power data corresponding to the current first time section and the second historical wind power data corresponding to each second time section in the current total reference wind power time series sample, select the second time section with the smallest Euclidean distance as the target second time section, correct the current first time section according to the second historical wind power data corresponding to the target second time section, obtain a corrected first time section, and remove the target second time section from the current total reference wind power time series sample;

[0185] In schematic form, the wind power time series samples to be corrected are generated based on the historical wind power data of each wind farm in month j using the traditional Markov Chain Monte Carlo (MCMC) algorithm. For example: Assume that there are t in month j j Time section, the first time section of the wind power time series sample to be corrected The value of Total reference wind power time series sample The value of the lth second time section is Based on this, traverse each first time section;

[0186] Specifically, for the current first time section, the Euclidean distance between it and each second time section is calculated, and by comparing all the calculated Euclidean distances, the second time section with the smallest Euclidean distance value is selected and determined as the target second time section, wherein the Euclidean distance The specific calculation formula is as follows:

[0187]

[0188] in, represents the l1th second time section; represents the l2th first time section, that is, the current first time section;

[0189] In a preferred embodiment, the correcting the current first time section according to the second historical wind power data corresponding to the target second time section includes:

[0190] Replacing the first historical wind power data corresponding to the current first time section with the second historical wind power data corresponding to the target second time section;

[0191] Specifically, after determining the target second time section, the first historical wind power data corresponding to the current first time section is replaced with the second historical wind power data corresponding to the target second time section, and after the replacement is completed, the target second time section is deleted to reduce data redundancy and speed up data analysis;

[0192] Thus, each first time section is traversed according to the above correction process, thereby obtaining all corrected first time sections.

[0193] S5. Integrate all corrected first time sections to obtain historical wind power time series samples that meet the spatiotemporal correlation;

[0194] Indicatively, all the corrected first time sections are integrated to obtain the historical wind power time series samples that meet the spatiotemporal correlation in the jth month; finally, the monthly historical wind power time series samples are merged to obtain the annual historical wind power time series samples;

[0195] Specifically, Figure 2 Take this as an example to illustrate the overall process of this embodiment:

[0196] Obtain historical wind power data for each wind farm in each month of a certain year;

[0197] For the historical wind power data of the mth month, first perform abnormal data processing and data normalization on the historical wind power data of the month to obtain normalized historical wind power data, then filter out zero elements in the normalized historical wind power data, and based on the number of filtered zero elements and combined with the constructed optimization model, determine the number of samples to be generated in each third equidistant interval, and after determining the number of samples, randomly generate samples to fill each third equidistant interval, thereby forming a historical wind power data set;

[0198] Calculating the probability density function p(x) and the cumulative distribution function F(x) of the Gaussian mixture model of the mth historical wind power data set, and executing steps S201 to S208 based on the initial correlation coefficient matrix, the correction parameters, the Gaussian mixture model, and the corresponding cumulative distribution function until a parameter wind power time series sample is obtained;

[0199] Then, according to the parameter wind power time series sample, each first time section of the wind power time series sample to be corrected is corrected to obtain each corrected first time section, and all corrected first time sections are integrated to obtain the historical wind power time series sample;

[0200] Finally, the historical wind power time series samples of each month of the year are integrated to obtain the annual historical wind power time series samples.

[0201] See Figure 3 , is a schematic diagram of the historical wind power data of four offshore wind farms (Offshore Wind Farm 1, Offshore Wind Farm 2, Offshore Wind Farm 3 and Offshore Wind Farm 4) in a coastal province in 2020; among them, Figure 3 (a) is a schematic diagram of the historical wind power data of offshore wind farm 1 in 2020. Figure 3 (b) is a schematic diagram of the historical wind power data of offshore wind farm 2 in 2020. Figure 3 (c) is a schematic diagram of the historical wind power data of offshore wind farm 3 in 2020. Figure 3 (d) is a schematic diagram of historical wind power data of offshore wind farm 4 in 2020;

[0202] See also Figure 4 , is a schematic diagram of the historical wind power data of four offshore wind farms (Offshore Wind Farm 1, Offshore Wind Farm 2, Offshore Wind Farm 3 and Offshore Wind Farm 4) in a coastal province in 2021; among them, Figure 4 (a) is a schematic diagram of the historical wind power data of offshore wind farm 1 in 2021. Figure 4 (b) is a schematic diagram of the historical wind power data of offshore wind farm 2 in 2021. Figure 4 (c) is a schematic diagram of the historical wind power data of offshore wind farm 3 in 2021. Figure 4 (d) is a schematic diagram of historical wind power data of offshore wind farm 4 in 2021;

[0203] The experiment was conducted using the historical wind power data obtained from January 2020 and January 2021. Figure 5 , is a schematic diagram of the effect of the Gaussian mixture model obtained by smoothing the historical wind power data comprehensively obtained from each offshore wind farm in this embodiment, wherein, Figure 5 (a) is a schematic diagram of the effect of the Gaussian mixture model of offshore wind farm 1. Figure 5 (b) is a schematic diagram of the effect of the Gaussian mixture model of offshore wind farm 2. Figure 5 (c) is a schematic diagram of the effect of the Gaussian mixture model of offshore wind farm 3. Figure 5 (d) is a schematic diagram of the effect of the Gaussian mixture model of offshore wind farm 4. Figure 5It can be seen that after 0-element smoothing, the Gaussian mixture model can better fit the probability information of the original sample, and there is no Gaussian mixture model spike caused by the mutation amount.

[0204] like Figures 6 to 9 As shown in FIG, it is a probability density comparison diagram between the historical wind power time series samples corrected in this embodiment and the total reference wind power time series samples, where: Figure 6 This is a comparison chart of the probability density of offshore wind farm 1 in the historical wind power time series sample. Figure 7 This is a comparison chart of the probability density of offshore wind farm 2 in the historical wind power time series sample. Figure 8 This is a comparison chart of the probability density of offshore wind farm 3 in the historical wind power time series sample. Figure 9 The probability density comparison diagram of offshore wind farm 4 in the historical wind power time series sample is shown in Figure 2. Figure 6 、 Figure 7 、 Figure 8 and Figure 9 It can be seen that the corrected historical wind power time series samples have good similarity with the total reference wind power time series samples.

[0205] like Figure 10 As shown in the figure, the Spearman correlation (spatial correlation) between the historical powers corresponding to the four wind farms, Offshore Wind Farm 1, Offshore Wind Farm 2, Offshore Wind Farm 3 and Offshore Wind Farm 4, in the corrected historical wind power time series samples is satisfied. Figure 10 The first row from left to right represents offshore wind farm 1, offshore wind farm 2, offshore wind farm 3 and offshore wind farm 4. Figure 10 The first column from left to right represents offshore wind farm 1, offshore wind farm 2, offshore wind farm 3 and offshore wind farm 4;

[0206] like Figure 11 As shown in the figure, the Spearman correlation (spatial correlation) between the historical powers corresponding to the four wind farms, Offshore Wind Farm 1, Offshore Wind Farm 2, Offshore Wind Farm 3, and Offshore Wind Farm 4, in the total reference wind power time series sample is satisfied. Figure 11 The first row from left to right represents offshore wind farm 1, offshore wind farm 2, offshore wind farm 3 and offshore wind farm 4. Figure 11 The first column from left to right represents offshore wind farm 1, offshore wind farm 2, offshore wind farm 3 and offshore wind farm 4;

[0207] Comparison Figure 10 and Figure 11 It was later found that the spatial correlation of the historical wind power time series samples obtained in this embodiment is very close to the spatial correlation of the total reference wind power time series samples, indicating that the historical wind power time series samples obtained in this embodiment have strong spatial correlation.

[0208] Wind Farm 1 Wind Farm 2 Wind Farm 3 Wind Farm 4 Error mean 0.033 0.0291 0.0267 0.0463

[0209] Table 1

[0210] Refer to Table 1, which shows the average error of the state transfer matrix of the historical wind power time series samples compared with the total reference wind power time series samples. It can be seen from Table 1 that the error of the state transfer matrix of the historical wind power time series samples is small, which proves that the historical wind power time series samples generated by the method of this embodiment have strong time correlation.

[0211] See Figure 12 , is the load power curve of the normalized annual historical wind power time series sample; see Figure 13 , is the wind power curve of offshore wind farm 1 in the normalized annual historical wind power time series sample; see Figure 14 , is the wind power curve of offshore wind farm 2 in the normalized annual historical wind power time series sample; see Figure 15 , is the wind power curve of offshore wind farm 3 in the normalized annual historical wind power time series sample; see Figure 16 , is the wind power curve of offshore wind farm 4 in the normalized annual historical wind power time series sample.

[0212] See Figure 17 , is a wind power time series sample correction device taking into account time and space correlation provided by one embodiment of the present invention, comprising: a data acquisition module, a reference wind power time series sample determination module, a total reference wind power time series sample determination module, a correction module, and a historical wind power time series sample generation module;

[0213] The data acquisition module is used to acquire correction parameters, time series samples of wind power to be corrected, and historical wind power data sets of each wind farm; wherein the time series samples of wind power to be corrected include a plurality of first time sections;

[0214] The reference wind power time series sample determination module is configured to fit the historical wind power data set of each wind farm to obtain a Gaussian mixture model for characterizing the distribution characteristics of the time series output curve, and calculate the reference wind power time series sample based on the initial correlation coefficient matrix, the correction parameters, and the Gaussian mixture model; wherein the initial correlation coefficient matrix is ​​calculated using the historical wind power data set;

[0215] The total reference wind power time series sample determination module is used to integrate the reference wind power time series samples of each wind farm to obtain the total reference wind power time series sample;

[0216] The correction module is used to traverse each first time section, and during each traversal, calculate the Euclidean distance between the first historical wind power data corresponding to the current first time section and the second historical wind power data corresponding to each second time section in the current total reference wind power time series sample, select the second time section with the smallest Euclidean distance as the target second time section, correct the current first time section according to the second historical wind power data corresponding to the target second time section, obtain a corrected first time section, and remove the target second time section from the current total reference wind power time series sample;

[0217] The historical wind power time series sample generation module is used to integrate all the corrected first time sections to obtain historical wind power time series samples that meet the time-space correlation.

[0218] See Figure 18 , an embodiment of the present application further provides a terminal device, including:

[0219] one or more processors;

[0220] a memory, coupled to the processor, for storing one or more programs;

[0221] When the one or more programs are executed by the one or more processors, the one or more processors implement the wind power time series sample correction method taking into account the temporal and spatial correlation as described above.

[0222] The processor is used to control the overall operation of the terminal device to complete all or part of the steps of the wind power time series sample correction method taking into account the temporal and spatial correlation. The memory is used to store various types of data to support the operation of the terminal device. For example, these data may include instructions for any application or method used to operate on the terminal device, as well as application-related data. The memory can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk.

[0223] In an exemplary embodiment, the terminal device can be implemented by one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components, and is used to execute the wind power timing sample correction method taking into account the temporal and spatial correlation as described in any of the above embodiments, and achieve the same technical effect as the above method.

[0224] In another exemplary embodiment, a computer-readable storage medium including a computer program is further provided. When executed by a processor, the computer program implements the steps of the wind power time series sample correction method taking into account spatiotemporal correlation as described in any of the above-mentioned embodiments. For example, the computer-readable storage medium may be the aforementioned memory including the computer program. The computer program may be executed by a processor of a terminal device to implement the wind power time series sample correction method taking into account spatiotemporal correlation as described in any of the above-mentioned embodiments, and achieve the same technical effects as the above-mentioned methods.

[0225] The above is a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A wind power time series sample correction method taking into account temporal and spatial correlation, characterized in that: include: Acquire correction parameters, time series samples of wind power to be corrected, and historical wind power data sets of each wind farm; wherein the time series samples of wind power to be corrected include a plurality of first time sections; For each wind farm, fitting the historical wind power data set of the wind farm to obtain a Gaussian mixture model for characterizing the distribution characteristics of the time-series output curve, and calculating a reference wind power time-series sample based on an initial correlation coefficient matrix, the correction parameters, and the Gaussian mixture model; wherein the initial correlation coefficient matrix is ​​calculated using the historical wind power data set; Integrate the reference wind power time series samples of each wind farm to obtain the total reference wind power time series samples; Traversing each first time section, during each traversal, calculating the Euclidean distance between the first historical wind power data corresponding to the current first time section and the second historical wind power data corresponding to each second time section in the current total reference wind power time series sample, selecting the second time section with the smallest Euclidean distance as the target second time section, correcting the current first time section according to the second historical wind power data corresponding to the target second time section to obtain a corrected first time section, and removing the target second time section from the current total reference wind power time series sample; By integrating all the corrected first time sections, we can obtain the historical wind power time series samples that meet the spatiotemporal correlation.

2. The wind power time series sample correction method taking into account the temporal and spatial correlation according to claim 1, characterized in that: The method for obtaining the historical wind power data set includes: Obtaining historical wind power data of the wind farm within a preset first time period; Normalizing the historical wind power data to obtain normalized historical wind power data; The historical wind power data is divided using the time characteristics of the normalized historical wind power data as a classification basis and using the length of a preset second time period as a separation scale to obtain a plurality of historical wind power data sets to be smoothed; wherein the length of the preset second time period is less than the length of the preset first time period; and the time periods covered by the historical wind power data sets to be smoothed do not overlap with each other; For each of the historical wind power data sets to be smoothed, performing data smoothing processing on the historical wind power data set to be smoothed to obtain a historical wind power data set after data smoothing processing; The historical wind power data set after the data smoothing process is used as the historical wind power data set.

3. The wind power time series sample correction method taking into account the temporal and spatial correlation according to claim 2, characterized in that: The step of performing data smoothing on the historical wind power data set to be smoothed to obtain the historical wind power data set after data smoothing includes: Dividing the historical wind power data set to be smoothed into a plurality of first equidistant intervals; Counting the total number of historical wind power data in each first equidistant interval to obtain a first total number; Identifying invalid data in each first equidistant interval, removing the invalid data from the corresponding first equidistant interval, and using the first equidistant interval after removing the invalid data as the second equidistant interval; Counting the total number of invalid data to obtain a second total number; According to the first total number, the second total number, and a preset optimization model, a number of samples that can be filled in a preset third equidistant interval is obtained; wherein the second equidistant interval and the third equidistant interval do not overlap with each other; Taking the number of fillable samples as the filling upper limit, randomly generating wind power data samples and filling them into the third equidistant interval to obtain the third equidistant interval after filling; Based on the third equidistant interval after filling and the second equidistant interval, a historical wind power data set after data smoothing is reconstructed; wherein the second equidistant interval and the third equidistant interval are located on the same reference coordinate axis.

4. The wind power time series sample correction method taking into account the temporal and spatial correlation according to claim 1, characterized in that: The step of calculating a reference wind power time series sample based on the initial correlation coefficient matrix, the correction parameter, and the Gaussian mixture model includes: Generating an initial reference wind power time series sample according to a probability density function corresponding to the Gaussian mixture model; Normalizing the initial wind power time series samples to obtain normalized reference wind power time series samples; The normalized reference wind power time series samples are corrected according to the initial correlation coefficient matrix, the correction parameters and the Gaussian mixture model to obtain final reference wind power time series samples.

5. The wind power time series sample correction method taking into account the temporal and spatial correlation according to claim 4, characterized in that: The correction parameters include the maximum number of iterations, convergence accuracy and change step size; The step of correcting the normalized reference wind power time series samples according to the initial correlation coefficient matrix, the correction parameter, and the Gaussian mixture model to obtain a final reference wind power time series sample includes: Repeating the iterative calculation operation according to the correction parameter, the initial correlation coefficient matrix and the Gaussian mixture model until the final reference wind power time series sample is obtained; The iterative calculation operation includes: Update the current number of iterations; Decomposing the current correlation coefficient matrix to obtain a lower triangular matrix; wherein, when performing the iterative calculation operation for the first time, the current correlation coefficient matrix is ​​set as the initial correlation coefficient matrix; According to the preset sample matrix obeying the standard normal distribution and the lower triangular matrix, a standard normal distribution sample obeying the correlation is calculated; Mapping the standard normal distribution samples to the original sample space that conforms to the multivariate normal distribution in the Gaussian mixture model to obtain candidate wind power time series samples; Calculating an approximate correlation coefficient matrix based on the candidate wind power time series samples; Calculating the difference between the approximate correlation coefficient matrix and the initial correlation coefficient matrix; Determine whether the difference is not less than the convergence accuracy, and determine whether the current number of iterations is less than the maximum number of iterations, If the judgment results are all yes, the current correlation coefficient matrix is ​​corrected according to the difference and the change step size to obtain a corrected correlation coefficient matrix, and the corrected correlation coefficient matrix is ​​used as the current correlation coefficient matrix for the next iterative calculation operation; Otherwise, the candidate wind power time series sample is used as the final reference wind power time series sample.

6. The wind power time series sample correction method taking into account the temporal and spatial correlation according to claim 5, characterized in that: Mapping the standard normal distribution samples to the original sample space conforming to the multivariate normal distribution in the Gaussian mixture model to obtain candidate wind power time series samples includes: Substituting each element in the standard normal distribution sample into the preset standard normal distribution cumulative distribution function, and calculating the probability value corresponding to each element; For each element in the standard normal distribution sample, according to the probability value of the element, an approximate value of the probability value is found in a preset table as the element of the alternative wind power time series sample; wherein, the preset table includes a number of equidistant points selected from the historical wind power data set and the cumulative distribution function value corresponding to each equidistant point; the cumulative distribution function value corresponding to each equidistant point is calculated by substituting each equidistant point into the Gaussian mixture model respectively.

7. The wind power time series sample correction method taking into account the temporal and spatial correlation according to claim 1, characterized in that: The correcting the current first time section according to the second historical wind power data corresponding to the target second time section includes: The first historical wind power data corresponding to the current first time section is replaced by the second historical wind power data corresponding to the target second time section.

8. A wind power time series sample correction device taking into account temporal and spatial correlation, characterized in that: include: Data acquisition module, reference wind power time series sample determination module, total reference wind power time series sample determination module, correction module and historical wind power time series sample generation module; The data acquisition module is used to acquire correction parameters, time series samples of wind power to be corrected, and historical wind power data sets of each wind farm; wherein the time series samples of wind power to be corrected include a plurality of first time sections; The reference wind power time series sample determination module is configured to fit the historical wind power data set of each wind farm to obtain a Gaussian mixture model for characterizing the distribution characteristics of the time series output curve, and calculate the reference wind power time series sample based on the initial correlation coefficient matrix, the correction parameters, and the Gaussian mixture model; wherein the initial correlation coefficient matrix is ​​calculated using the historical wind power data set; The total reference wind power time series sample determination module is used to integrate the reference wind power time series samples of each wind farm to obtain the total reference wind power time series sample; The correction module is used to traverse each first time section, and during each traversal, calculate the Euclidean distance between the first historical wind power data corresponding to the current first time section and the second historical wind power data corresponding to each second time section in the current total reference wind power time series sample, select the second time section with the smallest Euclidean distance as the target second time section, correct the current first time section according to the second historical wind power data corresponding to the target second time section, obtain a corrected first time section, and remove the target second time section from the current total reference wind power time series sample; The historical wind power time series sample generation module is used to integrate all the corrected first time sections to obtain historical wind power time series samples that meet the time-space correlation.

9. A terminal device, characterized in that: include: one or more processors; a memory, coupled to the processor, for storing one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the wind power time series sample correction method taking into account the temporal and spatial correlation as described in any one of claims 1 to 7.

10. A storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the wind power time series sample correction method taking into account the temporal and spatial correlation as described in any one of claims 1 to 7 is implemented.

Images