A power load fluctuation range prediction method and device

Abstract

The application relates to a power load fluctuation range prediction method and device, which comprises the following steps: obtaining a fuzzy matrix corresponding to a current time and a historical power load time sequence before the current time; reconstructing each row of fuzzy data in the fuzzy matrix in a phase space to obtain a phase space matrix corresponding to each row of fuzzy data; and determining a power load fluctuation range of a future period after the current time according to the phase space matrix corresponding to each row of fuzzy data; the phase space matrix corresponding to the historical power load time sequence is obtained, the phase space matrix is used to obtain a load fluctuation range at a prediction time, the chaos and uncertainty of the power load time sequence are considered, the prediction result is more accurate than that of the prior art, and strong support can be provided for subsequent power distribution planning.

Description

Technical Field

[0001] This invention relates to the field of power system load forecasting technology, and specifically to a method and apparatus for forecasting power load fluctuation range. Background Technology

[0002] With the development of big data and artificial intelligence technologies, more and more load forecasting technologies are being applied. However, the chaotic and variable nature of power load time series poses challenges to the application of accurate power load forecasting technologies. How to solve the prediction problem of chaotic power load series has become a research challenge. Existing research addresses this problem by combining linear and nonlinear models. That is, after modeling with a linear model, the prediction result of the linear model is subtracted from the original series to obtain a residual term containing nonlinear components; then, a nonlinear model is used to predict this residual term to obtain the final prediction result. This method often results in a residual series with high noise content and a low signal-to-noise ratio after subtracting the linear prediction component from the original time series. This makes it difficult for nonlinear methods to learn the characteristics of the nonlinear components, leading to low prediction accuracy.

[0003] The prediction results obtained by existing prediction models can only predict the fixed value of the load sequence at a certain moment. However, the power load time series is affected by many external factors and has strong uncertainty. It is reasonable and necessary to consider such uncertainties in load forecasting.

[0004] Therefore, to address the aforementioned problems, there is a need in this field for a method to predict the range of load fluctuations in chaotic time series. Summary of the Invention

[0005] To address the shortcomings of existing technologies, the present invention provides a method and apparatus for predicting power load fluctuation range. By obtaining the phase space matrix corresponding to the historical power load time series, the load fluctuation range at the prediction time is obtained using the phase space matrix. This method takes into account the chaos and uncertainty of the power load time series, and the prediction results are more accurate than those obtained by existing technologies, providing strong support for the formulation of subsequent power distribution plans.

[0006] The objective of this invention is achieved through the following technical solution:

[0007] This invention provides a method for predicting the range of power load fluctuations, the improvement of which is that the method includes:

[0008] Obtain the fuzzy matrix corresponding to the historical power load time series at the current time and before the current time;

[0009] The phase space is reconstructed for each row of fuzzy data in the fuzzy matrix to obtain the phase space matrix corresponding to each row of fuzzy data.

[0010] The range of power load fluctuations in future periods after the current moment is determined based on the phase space matrix corresponding to each row of fuzzy data.

[0011] Preferably, obtaining the fuzzy matrix corresponding to the historical power load time series at the current time and before the current time includes:

[0012] The historical power load time series is divided into multiple time windows based on a preset time length.

[0013] Obtain the fuzzy values ​​of data in each time window;

[0014] The fuzzy matrix corresponding to the historical power load time series is determined based on the fuzzy values ​​of the data in each time window.

[0015] Furthermore, obtaining the fuzzy values ​​of data in each time window includes:

[0016] The fuzzy value x of the k-th data in the n-th time window is determined by the following formula. n,k :

[0017]

[0018] In the formula, x′ n,k For the k-th data point in the n-th time window, m n a is the median of the data in the nth time window. n b is the minimum data in the nth time window. n This represents the maximum data value in the nth time window.

[0019] Furthermore, determining the fuzzy matrix corresponding to the historical power load time series based on the fuzzy values ​​of the data in each time window includes:

[0020] Based on the fuzzy values ​​of each data point in the nth time window, the element in the nth column of the fuzzy matrix corresponding to the historical power load time series is obtained using the following formula:

[0021]

[0022] In the formula, f n,1 f is the element in the first row and nth column of the fuzzy matrix. n,2 f is the element in the second row and nth column of the fuzzy matrix. n,3 x is the element in the third row and nth column of the fuzzy matrix. n,k Let x be the fuzzy value of the k-th data point in the n-th time window. n,mK′ is the fuzzy value of the median of the data in the nth time window, k∈[1,K], K is the number of data in the time window, K′ is the number of data before the median of the time window, n∈[1,N], N is the total number of time windows obtained by dividing the historical power load time series into units of preset time length;

[0023] The fuzzy matrix F corresponding to the historical power load time series is determined by the following formula:

[0024]

[0025] Preferably, the step of reconstructing the phase space of each row of fuzzy data in the fuzzy matrix to obtain the phase space matrix corresponding to each row of fuzzy data includes:

[0026] The embedding dimension of the phase space corresponding to each row of fuzzy data in the fuzzy matrix is ​​obtained by using the spurious neighbor method, and the delay time of the phase space corresponding to each row of fuzzy data in the fuzzy matrix is ​​obtained by using the autocorrelation method.

[0027] The phase space matrix corresponding to each row of fuzzy data is obtained based on the minimum embedding dimension among the embedding dimensions of the phase space corresponding to each row of fuzzy data and the minimum delay time among the delay times of the phase space corresponding to each row of fuzzy data.

[0028] Preferably, determining the power load fluctuation range for future periods after the current time based on the phase space matrix corresponding to each row of fuzzy data includes:

[0029] Substitute the phase space data in the Q-T+s column of the phase space matrix corresponding to the i-th row of fuzzy data into the pre-acquired i-th support vector machine model to obtain the last phase space prediction data in the Q+s column of the phase space matrix corresponding to the i-th row of fuzzy data.

[0030] The last phase space prediction data in the Q+s column of the phase space matrix corresponding to each row of fuzzy data is used as the output of the power load fluctuation range in the s-th future time period after the current time.

[0031] Where i∈(1,2,3), Q is the total number of columns in the phase space matrix, T is the prediction step size, and the time length of the future time period is the same as the preset time length.

[0032] Furthermore, the process of obtaining the pre-acquired i-th support vector machine model includes:

[0033] The i-th row of fuzzy data is used as the input layer sample, and the last phase space data in the q+T column of the phase space matrix corresponding to the i-th row of fuzzy data is used as the output layer sample. The pre-built support vector machine model is trained to obtain the pre-acquired i-th support vector machine model.

[0034] Where q∈[1,QT].

[0035] Furthermore, the prediction step size T is obtained as follows:

[0036] The prediction step size t of the phase space matrix corresponding to the i-th row of fuzzy data is determined by the following formula. i :

[0037]

[0038] In the formula, λ i Let Lyapunov exponent be the phase space matrix corresponding to the i-th row of fuzzy data;

[0039] The smallest prediction step size among the prediction step sizes of the phase space matrix corresponding to each row of fuzzy data is selected as the prediction step size T.

[0040] Based on the same inventive concept, the present invention also provides a power load fluctuation range prediction device, wherein the improvement is that the device includes:

[0041] The first acquisition unit is used to acquire the fuzzy matrix corresponding to the historical power load time series at the current time and before the current time.

[0042] The second acquisition unit is used to reconstruct the phase space of each row of fuzzy data in the fuzzy matrix and obtain the phase space matrix corresponding to each row of fuzzy data.

[0043] The prediction unit is used to determine the range of power load fluctuations in the future period after the current moment based on the phase space matrix corresponding to each row of fuzzy data.

[0044] Compared with the closest existing technology, the present invention has the following advantages:

[0045] This invention provides a method and apparatus for predicting power load fluctuation range. The method involves obtaining a fuzzy matrix corresponding to the historical power load time series at the current time and before the current time; reconstructing the phase space of each row of fuzzy data in the fuzzy matrix to obtain the phase space matrix corresponding to each row of fuzzy data; and determining the power load fluctuation range for future periods after the current time based on the phase space matrix corresponding to each row of fuzzy data. This invention smooths the sequence fluctuations and reduces the randomness of the sequence by fuzzifying the sequence during the process of obtaining the phase space matrix corresponding to the historical power load time series, making the subsequent model training process more reliable. Using the phase space matrix data for model training and prediction preserves the characteristics of the nonlinear components in the original sequence, taking into account the chaos and uncertainty of the power load time series. The prediction results are more accurate than those obtained by existing technologies, providing strong support for the formulation of subsequent power distribution plans. Attached Figure Description

[0046] Figure 1 This is a flowchart of the power load fluctuation range prediction method of the present invention;

[0047] Figure 2 This is a schematic diagram of the power load fluctuation range prediction device of the present invention. Detailed Implementation

[0048] The specific embodiments of the present invention will be further described in detail below with reference to the accompanying drawings.

[0049] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0050] This invention provides a method for predicting the range of power load fluctuations, such as... Figure 1 As shown, the method includes:

[0051] Obtain the fuzzy matrix corresponding to the historical power load time series at the current time and before the current time;

[0052] The phase space is reconstructed for each row of fuzzy data in the fuzzy matrix to obtain the phase space matrix corresponding to each row of fuzzy data.

[0053] The range of power load fluctuations in future periods after the current moment is determined based on the phase space matrix corresponding to each row of fuzzy data.

[0054] To more clearly illustrate the purpose of this invention, the method of this invention will be described below with reference to specific embodiments.

[0055] To make the forecasts more accurate, historical power load time series can be obtained by time category. The categories can be divided by date index (weekdays, weekends, and holidays) and seasonal index (four quarters), resulting in 12 categories. In this classification, the historical moments in the series and the predicted future periods are not consecutive natural moments.

[0056] In an embodiment of the present invention, obtaining the fuzzy matrix corresponding to the historical power load time series at the current time and before the current time includes:

[0057] The historical power load time series is divided into multiple time windows based on a preset time length.

[0058] Obtain the fuzzy values ​​of data in each time window;

[0059] The fuzzy matrix corresponding to the historical power load time series is determined based on the fuzzy values ​​of the data in each time window.

[0060] Specifically, the above-mentioned acquisition of fuzzy values ​​of data in each time window includes:

[0061] The fuzzy value x of the k-th data in the n-th time window is determined by the following formula. n,k :

[0062]

[0063] In the formula, x′ n,k For the k-th data point in the n-th time window, m n a is the median of the data in the nth time window. n b is the minimum data in the nth time window. n This represents the maximum data value in the nth time window.

[0064] Specifically, the above-mentioned determination of the fuzzy matrix corresponding to the historical power load time series based on the fuzzy values ​​of the data in each time window includes:

[0065] Based on the fuzzy values ​​of each data point in the nth time window, the element in the nth column of the fuzzy matrix corresponding to the historical power load time series is obtained using the following formula:

[0066]

[0067] In the formula, f n,1 f is the element in the first row and nth column of the fuzzy matrix. n,2 f is the element in the second row and nth column of the fuzzy matrix. n,3 x is the element in the third row and nth column of the fuzzy matrix. n,k Let x be the fuzzy value of the k-th data point in the n-th time window. n,m K′ is the fuzzy value of the median of the data in the nth time window, k∈[1,K], K is the number of data in the time window, K′ is the number of data before the median of the time window, n∈[1,N], N is the total number of time windows obtained by dividing the historical power load time series into units of preset time length;

[0068] The fuzzy matrix F corresponding to the historical power load time series is determined by the following formula:

[0069]

[0070] In an embodiment of the present invention, the above-mentioned phase space reconstruction of each row of fuzzy data in the fuzzy matrix to obtain the phase space matrix corresponding to each row of fuzzy data includes:

[0071] The embedding dimension of the phase space corresponding to each row of fuzzy data in the fuzzy matrix is ​​obtained by using the spurious neighbor method, and the delay time of the phase space corresponding to each row of fuzzy data in the fuzzy matrix is ​​obtained by using the autocorrelation method.

[0072] The phase space matrix corresponding to each row of fuzzy data is obtained based on the minimum embedding dimension among the embedding dimensions of the phase space corresponding to each row of fuzzy data and the minimum delay time among the delay times of the phase space corresponding to each row of fuzzy data.

[0073] The method for obtaining the aforementioned delay time τ is as follows:

[0074] Taking the first row of the fuzzy matrix as an example, we calculate the average mutual information I(τ) between the observations of the sequence at time n and time n+τ.

[0075]

[0076] In the formula, P(x n ), P(x n+τ ), P(x n ,x n+τ ) are all probabilities, x n This is the nth data point to be reconstructed.

[0077] Choosing τ as the delay time when I(τ) reaches its first local minimum ensures that x n With x n+τ It has a certain degree of independence but is not completely unrelated.

[0078] The method for obtaining the embedding dimension M is as follows:

[0079] Let x n The nearest neighbor is x η(n) When the embedding dimension is m, the nearest neighbor is represented as:

[0080]

[0081] When the embedding dimension increases to m+1, the nearest neighbor becomes:

[0082]

[0083] The principle for selecting parameter M is to make the proportion of spurious nearest neighbors (i.e., points that are far apart in the phase space due to the small embedding dimension but are projected into the lower dimension and thus become closer to each other in the reconstructed phase space) close to 0, and l is a constant.

[0084] like Compare If the value is greater than the preset value, then m is considered not to meet the requirement. In this case, the value of m needs to be increased, and the first value that satisfies this condition is selected. and Similar values ​​for m are used as the embedding dimension;

[0085] The minimum value of the delay time corresponding to the three rows in the fuzzy matrix and the minimum value of the embedding dimension are taken as the final result.

[0086] In an embodiment of the present invention, determining the range of power load fluctuations in the future period after the current moment based on the phase space matrix corresponding to each row of fuzzy data includes:

[0087] Substitute the phase space data in the Q-T+s column of the phase space matrix corresponding to the i-th row of fuzzy data into the pre-acquired i-th support vector machine model to obtain the last phase space prediction data in the Q+s column of the phase space matrix corresponding to the i-th row of fuzzy data.

[0088] The last phase space prediction data in the Q+s column of the phase space matrix corresponding to each row of fuzzy data is used as the output of the power load fluctuation range in the s-th future time period after the current time.

[0089] Where i∈(1,2,3), Q is the total number of columns in the phase space matrix, T is the prediction step size, and the time length of the future time period is the same as the preset time length.

[0090] Furthermore, the process of obtaining the pre-acquired i-th support vector machine model includes:

[0091] The i-th row of fuzzy data is used as the input layer sample, and the last phase space data in the q+T column of the phase space matrix corresponding to the i-th row of fuzzy data is used as the output layer sample. The pre-built support vector machine model is trained to obtain the pre-acquired i-th support vector machine model.

[0092] Where q∈[1,QT].

[0093] Furthermore, the prediction step size T is obtained as follows:

[0094] The prediction step size t of the phase space matrix corresponding to the i-th row of fuzzy data is determined by the following formula. i :

[0095]

[0096] In the formula, λ i Let Lyapunov exponent be the phase space matrix corresponding to the i-th row of fuzzy data;

[0097] The smallest prediction step size among the prediction step sizes of the phase space matrix corresponding to each row of fuzzy data is selected as the prediction step size T.

[0098] In an embodiment of the present invention, taking the first row of the fuzzy matrix as an example, the original time series is [x1, x2, ..., x...]. N After phase space reconstruction, the resulting phase space matrix is:

[0099] X = [X1, X2, ..., X q ,...,X Q ]

[0100] In the formula, X q =[x q ,x q+τ ,...,x q+(M-1)τ ] T (q = 1, 2, ..., Q), X q+T =[x q+k ,x q+k+τ ,…,x q+k+(M-1)τ ] T Q = N - (M - 1)τ, where M is the embedding dimension of the phase space, τ is the delay time of the phase space, and x q+(M-1)τ This represents the last phase space data in the q-th column of the phase space matrix X.

[0101] The principles for selecting relevant data for the above model training and prediction are as follows:

[0102] The principle analysis is as follows:

[0103] Determine the training data and label data for training the SVM.

[0104] Taking the phase space matrix obtained by reconstructing the phase space of the first row of the fuzzy matrix as an example, when the phase point X in the phase space is... q As training data, phase point X q+T As label data:

[0105] X q+T =[x q+T ,x q+T+τ ,…,x q+T+(m-1)τ ,…,x q+T+(M-1)τ ] T

[0106] After training the model, X q+T =ΦX q A state transition matrix can be obtained. Based on this state transition matrix,

[0107] We can obtain:

[0108]

[0109] ...

[0111]

[0112]

[0113] in, The predicted values ​​for the next T phase points

[0114] Since the first m-1 numbers of the label vector are known each time the state transition matrix is ​​calculated (through x...), q+(m-1)τ As can be seen from the subscript and Q = N - (m - 1)τ, the training process can be simplified, requiring only the training of the m-th row of the state transition matrix.

[0115] Set the label data to x q+T+(m-1)τ Train the state transition vector Ψ such that x q+T+(m-1)τ =ΨX q+T ,at this time:

[0116]

[0117] ...

[0119]

[0120]

[0121] in, For the original time series [x1,x2,...,x] N [The k predicted values ​​for the future.]

[0122] The specific training process of the above model includes: using the particle swarm optimization algorithm to obtain the model parameters of the support vector machine model, wherein the training error of the SVM model, i.e., the k-fold cross-validation error, is used as the fitness function of the particle swarm optimization algorithm.

[0123] The phase point X in phase space q As input layer training data, the phase point X q+T Used as training data for the output layer.

[0124] Furthermore, in a preferred embodiment, when training the model, other columns of data in the phase space matrix can be selected as test data according to conventional training methods to verify whether the model is overfitting. If overfitting occurs, the parameters of the phase space matrix are modified.

[0125] Based on the same inventive concept, the present invention also provides a power load fluctuation range prediction device, such as... Figure 2 As shown, the device includes:

[0126] The first acquisition unit is used to acquire the fuzzy matrix corresponding to the historical power load time series at the current time and before the current time.

[0127] The second acquisition unit is used to reconstruct the phase space of each row of fuzzy data in the fuzzy matrix and obtain the phase space matrix corresponding to each row of fuzzy data.

[0128] The prediction unit is used to determine the range of power load fluctuations in the future period after the current moment based on the phase space matrix corresponding to each row of fuzzy data.

[0129] Preferably, obtaining the fuzzy matrix corresponding to the historical power load time series at the current time and before the current time includes:

[0130] The historical power load time series is divided into multiple time windows based on a preset time length.

[0131] Obtain the fuzzy values ​​of data in each time window;

[0132] The fuzzy matrix corresponding to the historical power load time series is determined based on the fuzzy values ​​of the data in each time window.

[0133] Furthermore, obtaining the fuzzy values ​​of data in each time window includes:

[0134] The fuzzy value x of the k-th data in the n-th time window is determined by the following formula. n,k :

[0135]

[0136] In the formula, x′ n,k For the k-th data point in the n-th time window, m n a is the median of the data in the nth time window. n b is the minimum data in the nth time window. n This represents the maximum data value in the nth time window.

[0137] Furthermore, determining the fuzzy matrix corresponding to the historical power load time series based on the fuzzy values ​​of the data in each time window includes:

[0138] Based on the fuzzy values ​​of each data point in the nth time window, the element in the nth column of the fuzzy matrix corresponding to the historical power load time series is obtained using the following formula:

[0139]

[0140] In the formula, f n,1 f is the element in the first row and nth column of the fuzzy matrix. n,2 f is the element in the second row and nth column of the fuzzy matrix. n,3x is the element in the third row and nth column of the fuzzy matrix. n,k Let x be the fuzzy value of the k-th data point in the n-th time window. n,m K′ is the fuzzy value of the median of the data in the nth time window, k∈[1,K], K is the number of data in the time window, K′ is the number of data before the median of the time window, n∈[1,N], N is the total number of time windows obtained by dividing the historical power load time series into units of preset time length;

[0141] The fuzzy matrix F corresponding to the historical power load time series is determined by the following formula:

[0142]

[0143] Preferably, the step of reconstructing the phase space of each row of fuzzy data in the fuzzy matrix to obtain the phase space matrix corresponding to each row of fuzzy data includes:

[0144] The embedding dimension of the phase space corresponding to each row of fuzzy data in the fuzzy matrix is ​​obtained by using the spurious neighbor method, and the delay time of the phase space corresponding to each row of fuzzy data in the fuzzy matrix is ​​obtained by using the autocorrelation method.

[0145] The phase space matrix corresponding to each row of fuzzy data is obtained based on the minimum embedding dimension among the embedding dimensions of the phase space corresponding to each row of fuzzy data and the minimum delay time among the delay times of the phase space corresponding to each row of fuzzy data.

[0146] Preferably, determining the power load fluctuation range for future periods after the current time based on the phase space matrix corresponding to each row of fuzzy data includes:

[0147] Substitute the phase space data in the Q-T+s column of the phase space matrix corresponding to the i-th row of fuzzy data into the pre-acquired i-th support vector machine model to obtain the last phase space prediction data in the Q+s column of the phase space matrix corresponding to the i-th row of fuzzy data.

[0148] The last phase space prediction data in the Q+s column of the phase space matrix corresponding to each row of fuzzy data is used as the output of the power load fluctuation range in the s-th future time period after the current time.

[0149] Where i∈(1,2,3), Q is the total number of columns in the phase space matrix, T is the prediction step size, and the time length of the future time period is the same as the preset time length.

[0150] Furthermore, the process of obtaining the pre-acquired i-th support vector machine model includes:

[0151] The i-th row of fuzzy data is used as the input layer sample, and the last phase space data in the q+T column of the phase space matrix corresponding to the i-th row of fuzzy data is used as the output layer sample. The pre-built support vector machine model is trained to obtain the pre-acquired i-th support vector machine model.

[0152] Where q∈[1,QT].

[0153] Furthermore, the prediction step size T is obtained as follows:

[0154] The prediction step size t of the phase space matrix corresponding to the i-th row of fuzzy data is determined by the following formula. i :

[0155]

[0156] In the formula, λ i Let Lyapunov exponent be the phase space matrix corresponding to the i-th row of fuzzy data;

[0157] The smallest prediction step size among the prediction step sizes of the phase space matrix corresponding to each row of fuzzy data is selected as the prediction step size T.

[0158] In summary, the present invention provides a method and apparatus for predicting power load fluctuation range. It acquires the phase space matrix and prediction step size corresponding to the historical power load time series at the current time and before the current time; and determines the power load fluctuation range for future periods after the current time based on the phase space matrix and prediction step size. The present invention performs fuzzification processing on the sequence during the acquisition of the phase space matrix corresponding to the historical power load time series, smoothing the sequence fluctuations and reducing the randomness of the sequence, making the subsequent model training process more reliable. Using the phase space matrix data for model training and prediction preserves the characteristics of the nonlinear components in the original sequence, taking into account the chaos and uncertainty of the power load time series. The prediction results are more accurate than those obtained by existing technologies, providing strong support for the formulation of subsequent power distribution plans. In this embodiment, the particle swarm optimization algorithm is used to obtain the model parameters, reducing the computational load and ensuring the accuracy of the parameters, resulting in a more accurate model. Furthermore, obtaining the historical power load time series by category can further enhance the prediction accuracy.

[0159] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0160] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0161] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0162] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0163] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A method for predicting the range of power load fluctuations, characterized in that, The method includes: Obtain the fuzzy matrix corresponding to the historical power load time series at the current time and before the current time; The phase space is reconstructed for each row of fuzzy data in the fuzzy matrix to obtain the phase space matrix corresponding to each row of fuzzy data. The range of power load fluctuations in future periods after the current moment is determined based on the phase space matrix corresponding to each row of fuzzy data. The step of obtaining the fuzzy matrix corresponding to the historical power load time series at the current time and before the current time includes: The historical power load time series is divided into multiple time windows based on a preset time length. Obtain the fuzzy values ​​of data in each time window; The fuzzy matrix corresponding to the historical power load time series is determined based on the fuzzy values ​​of the data in each time window, including: Based on the fuzzy values ​​of each data point in the nth time window, the element in the nth column of the fuzzy matrix corresponding to the historical power load time series is obtained using the following formula: In the formula, The element in the first row and nth column of the fuzzy matrix. The element in the second row and nth column of the fuzzy matrix. The element in the third row and nth column of the fuzzy matrix. For the fuzzy value of the k-th data point in the n-th time window, Let be the fuzzy value of the median of the data in the nth time window. K is the number of data points in the time window. This represents the number of data points preceding the median of the time window. N is the total number of time windows obtained by dividing the historical power load time series into units of preset time length; The fuzzy matrix F corresponding to the historical power load time series is determined by the following formula: ； The step of reconstructing the phase space of each row of fuzzy data in the fuzzy matrix to obtain the phase space matrix corresponding to each row of fuzzy data includes: The embedding dimension of the phase space corresponding to each row of fuzzy data in the fuzzy matrix is ​​obtained by using the spurious neighbor method, and the delay time of the phase space corresponding to each row of fuzzy data in the fuzzy matrix is ​​obtained by using the autocorrelation method. The phase space matrix corresponding to each row of fuzzy data is obtained based on the minimum embedding dimension among the embedding dimensions of the phase space corresponding to each row of fuzzy data and the minimum delay time among the delay times of the phase space corresponding to each row of fuzzy data.

2. The method as described in claim 1, characterized in that, The process of obtaining fuzzy values ​​of data in each time window includes: The fuzzy value of the k-th data point in the n-th time window is determined by the following formula. : In the formula, For the k-th data point in the n-th time window, The median of the data in the nth time window. The minimum data for the nth time window. This represents the maximum data value in the nth time window.

3. The method as described in claim 1, characterized in that, The step of determining the power load fluctuation range for future periods after the current time based on the phase space matrix corresponding to each row of fuzzy data includes: Substitute the phase space data in the Q-T+s column of the phase space matrix corresponding to the i-th row of fuzzy data into the pre-acquired i-th support vector machine model to obtain the last phase space prediction data in the Q+s column of the phase space matrix corresponding to the i-th row of fuzzy data. The last phase space prediction data in the Q+s column of the phase space matrix corresponding to each row of fuzzy data is used as the output of the power load fluctuation range in the s-th future time period after the current time. in, Q is the total number of columns in the phase space matrix, T is the prediction step size, and the duration of the future time period is the same as the preset time length.

4. The method as described in claim 3, characterized in that, The process of obtaining the pre-acquired i-th support vector machine model includes: The i-th row of fuzzy data is used as the input layer sample, and the last phase space data in the q+T column of the phase space matrix corresponding to the i-th row of fuzzy data is used as the output layer sample. The pre-built support vector machine model is trained to obtain the pre-acquired i-th support vector machine model. in, .

5. The method as described in claim 3, characterized in that, The prediction step size T is obtained using the following method: The prediction step size of the phase space matrix corresponding to the i-th row of fuzzy data is determined by the following formula. : In the formula, Let Lyapunov exponent be the phase space matrix corresponding to the i-th row of fuzzy data; The smallest prediction step size among the prediction step sizes of the phase space matrix corresponding to each row of fuzzy data is selected as the prediction step size T.

6. A power load fluctuation range prediction device, used to implement the method as described in any one of claims 1-5, characterized in that, The device includes: The first acquisition unit is used to acquire the fuzzy matrix corresponding to the historical power load time series at the current time and before the current time. The second acquisition unit is used to reconstruct the phase space of each row of fuzzy data in the fuzzy matrix and obtain the phase space matrix corresponding to each row of fuzzy data. The prediction unit is used to determine the range of power load fluctuations in the future period after the current moment based on the phase space matrix corresponding to each row of fuzzy data.

Images