Electric energy quality real-time evaluation method and system based on cooperative game theory and dynamic entropy variable weight

Abstract

The invention discloses an electric energy quality real-time evaluation method and system based on a cooperative game theory and a dynamic entropy variable weight. The method comprises the following steps: collecting time sequence data of an electric energy quality index; performing window sliding standardization on the time sequence data to obtain standardized data; obtaining entropy values of the indexes and a correlation coefficient matrix among the indexes according to the standardized data; obtaining a Shapley value of the index according to the entropy value and the correlation coefficient matrix in combination with a cooperative game theory; obtaining a constant weight vector of the index according to the Shapley value; calculating a conflict entropy weight of the index and obtaining an expert subjective weight of the index; correcting the normal weight vector according to the conflict entropy weight and the expert subjective weight to obtain an index correction weight; normalizing the correction weights of the indexes to obtain correction normal weight vectors of the indexes; performing dynamic weight change on the corrected weight vector through a variable weight factor of grade constraint to obtain a first weight vector of a preset power quality index; and performing real-time evaluation on the power quality of the power distribution network node according to the first weight vector.

Description

Technical Field

[0001] The present invention relates to the field of power quality index evaluation of a new type of distribution network, and particularly to a real-time power quality evaluation method and system based on cooperative game theory and dynamic entropy value variable weight. Background Art

[0002] With the continuous increase in the penetration rate of renewable energy in the power grid, the power quality problems faced by the power grid have become increasingly prominent. Typical renewable energy sources such as wind and light usually adopt the maximum power point tracking control strategy for power generation. The uncertain natural environmental factors such as climate and geography on which they depend are more likely to cause sudden changes in output power, which in turn leads to power imbalance in the power grid, seriously affecting the frequency and voltage stability of the power grid. In addition, the inherent intermittency of new energy power generation and the harmonic pollution caused by the grid connection of power electronic devices have led to a complex situation where power quality problems such as voltage fluctuations and frequency deviations in the distribution network are intertwined in space and time. Therefore, establishing a power quality evaluation system that adapts to the dynamic characteristics of the new power system has become the key foundation for ensuring the energy transformation and the high-quality development of the power market.

[0003] The comprehensive evaluation of power quality covers multiple key indicators, mainly including parameters such as voltage, harmonics, and frequency. Since the meanings and dimensions of such technical indicators are different, when comprehensively evaluating power quality through power quality data, the original data of each indicator is usually standardized, and through data mining and analysis methods, an evaluation result that can comprehensively reflect the quality of power quality is obtained. This evaluation method can more comprehensively reveal the actual operating conditions of the power grid through data, help discover potential problems, and provide guidance for taking corresponding improvement measures. This evaluation method through data mining and analysis has high requirements for the effective information in the data, and the information equivalent in the data will directly affect the rationality of the evaluation result. Among them, using the analysis of data entropy value as the basis for index weighting has been studied by many workers. As a method of weighting by analyzing the differences between data, the entropy weight method is suitable for multi-index parameter evaluation. Some scholars have used the entropy weight method to construct static weights based on the data dispersion of power quality indicators, and quantify the index differences through information entropy to evaluate the comprehensive level of power quality. However, the traditional entropy weight method assumes that the indicators are independent of each other and the weights are fixed, without considering the dynamic cooperation or conflict effects of indicators such as voltage and harmonics during peak and valley periods of electricity consumption, resulting in a deviation between the evaluation result and the actual working conditions. In response to this, some scholars have proposed the dynamic entropy weight method, which introduces a sliding window to update the weights to respond to data changes. However, the dynamic entropy weight method still relies on global standardization processing, blurs the periodic fluctuation characteristics of the indicators, and does not quantify the deep influence of the game relationship between indicators on weight allocation.

[0004] Therefore, there is an urgent need for a new technical solution to solve the technical problem that when the traditional power quality assessment method of the distribution network is used to conduct real-time assessment on multi-index time series data, the assessment results are difficult to meet the requirements of universal objectivity and timeliness accuracy. Summary of the Invention

[0005] The present invention provides a real-time power quality assessment method and system based on cooperative game theory and dynamic entropy value variable weight, so as to solve the technical problem that when the traditional power quality assessment method of the distribution network is used to conduct real-time assessment on multi-index time series data, the assessment results are difficult to meet the requirements of universal objectivity and timeliness accuracy.

[0006] To achieve the above object, the present invention provides a real-time power quality assessment method based on cooperative game theory and dynamic entropy value variable weight, including:

[0007] Collect the time series data of the preset power quality indexes of the distribution network nodes; perform window sliding normalization on the time series data to obtain normalized data; obtain the entropy values of each index and the correlation coefficient matrix between the indexes according to the normalized data;

[0008] Obtain the Shapley value of each index according to the entropy value and the correlation coefficient matrix in combination with the cooperative game theory; obtain the constant weight vector of the index according to the Shapley value; calculate the conflict entropy weight of each index, and obtain the expert subjective weight of each index;

[0009] Modify the constant weight vector according to the conflict entropy weight and the expert subjective weight to obtain the index correction weight; normalize the index correction weight to obtain the corrected constant weight vector of the index;

[0010] Perform dynamic variable weight on the corrected constant weight vector through the variable weight factor under the level constraint to obtain the first weight vector of the preset power quality index; perform real-time assessment on the power quality of the distribution network nodes according to the first weight vector.

[0011] Preferably, performing window sliding normalization on the time series data to obtain normalized data includes:

[0012] Divide the window length and the sliding step according to the preset statistical period;

[0013] Confirm the initialization window at the starting point of the time series, move the window by the step length, discard the oldest data point, incorporate the latest data point, and normalize the power quality time series data of each window. The normalization includes:

[0014]

[0015] Among them, represents the normalized value of the s-th data point in the m-th window; x (m)(s) represents the s-th data value in the m-th window; N represents the length of a single window of power quality time series data, that is, the total number of sampling points is N; x (m) (k) represents the data value of the k-th data point in the m-th window; k represents the data point in the window, where m ≤ k ≤ m + N - 1.

[0016] Preferably, obtaining the entropy value of each index from the standardized data includes:

[0017] Perform normalization processing on the standardized data:

[0018]

[0019] After performing normalization processing on the standardized data of all M power quality indexes, integrate to obtain the data initialization matrix X (m) :

[0020]

[0021] Among them, represents the i-th normalized data of the j-th index in the m-th window; represents the standardized value of the first data point in the m-th window; represents the standardized value of the s-th data point in the m-th window; M represents the total number of power quality indexes;

[0022] The entropy value of the index includes:

[0023]

[0024] represents the proportion of the i-th sample value of the j-th index in the m-th window to this index:

[0025]

[0026] When At this time, is meaningless, so is corrected:

[0027]

[0028] Among them, ε is a correction factor. To avoid a huge error in the calculation of the entropy value, ε is taken as a minimum value.

[0029] Preferably, the correlation coefficient matrix between indexes includes:

[0030] Calculate the Pearson correlation coefficient r between each index ij And obtain the correlation coefficient matrix

[0031]

[0032] Among them, r ij represents the Pearson correlation coefficient between index i and index j; x n and y n respectively represent the nth normalized values of index i and index j in the mth time series window; and respectively represent the average normalized data of index i and index j in the mth time series window; represents the correlation coefficient between index i and index j in the mth time series window.

[0033] Preferably, obtaining the Shapley value of each index according to the entropy value and the correlation coefficient matrix in combination with the cooperative game theory includes:

[0034] Regarding the index as a participant in the cooperative game, define the dynamic revenue function v(S) of the index coalition S:

[0035]

[0036] Among them, S represents all non-overlapping sets that can be formed by all M indices, that is, the index coalition, with a total of 2 M -1; E j represents the entropy value of the jth index; t is the index of the time window, representing the time step of the dynamic model; τ represents the moment index within the time window; L represents the total window length; α is the time decay factor, which assigns higher weights to recent data through the time decay factor, α < 1; x j,τ is the normalized data of index j at the moment τ; β(S) is the sum of the absolute values of the correlation coefficients of all index pairs within the coalition S;

[0037] Then the Shapley value φ j of the index includes:

[0038]

[0039] Among them, A is the set of participants, representing all power quality indices; A\{j} represents the subset obtained by removing index j from set A; φ j represents the weight Shapley value of the jth index in the entire window; {j} represents the set formed by the jth index; v(S∪{j}) represents the value function of the coalition containing index j.

[0040] Preferably, obtaining the constant weight vector of the index according to the Shapley value includes:

[0041] The constant weight vector of the index includes:

[0042]

[0043] Among them, B j represents the weight of the j-th index in the entire window. Calculating the weight of each index yields the constant weight vector of the index.

[0044] Preferably, calculating the conflict entropy weight of each index includes:

[0045]

[0046] Among them, Q j is the conflict intensity of index j, P j is the conflict entropy weight, is the average entropy value of the j-th index within the time window; E j is the entropy value of the j-th index; Q c is the conflict intensity of index c; E c is the entropy value of the c-th index.

[0047] Preferably, the constant weight vector is corrected according to the conflict entropy weight and the expert subjective weight to obtain the corrected weight of the index; the corrected weight of the index is normalized to obtain the corrected constant weight vector of the index, including:

[0048] Adopt the idea of cooperative game to correct the constant weight vector with the expert subjective weight and the conflict entropy weight:

[0049] Calculate the consistency correlation coefficient L(h) of the expert subjective weight, the conflict entropy weight and the constant weight vector:

[0050]

[0051] Among them, W j (h) represents the h-th weight of index j, W(d_h) represents the optimal combined weight of d - 1 weights W(1),..., W(h - 1), W(h + 1),..., W(d) excluding W(h); d represents the total number of weight calculation methods; and respectively represent the average values of W j (h) and W j (d_h);

[0052] Solve for the corrected weight of the index W j ', including:

[0053]

[0054] Normalize the corrected weight of the index W j ' to obtain the corrected constant weight vector of the index W = (w1, w2,..., w M ).

[0055] Preferably, dynamically varying the modified normal weight vector by the variable weight factor under the rank constraint to obtain the first weight vector of the preset power quality index includes:

[0056] Taking the rank limit as the index for quantifying the deterioration degree of the index, the variable weight factor includes:

[0057] T(x) = e λV(x)

[0058] Among them, T(x) represents the variable weight factor of the index data x of the data point; λ represents the gain coefficient of the index variable weight factor, and the value of λ affects the variable weight benefit; V(x) is the rank membership function of the index data x of the data point, used to reflect the deterioration degree of the data x, including:

[0059]

[0060] Among them, x represents the index data of the data point in the time window; V g is the rank coefficient under the g-th rank of the index; [x g-1 , x g is the upper and lower rank limits of the g-th rank; x g represents the highest rank limit under the g-th rank of the index; x g-1 represents the highest rank limit under the (g - 1)-th rank of the index; V gmax represents the maximum rank coefficient of the index; x gmax represents the highest rank limit under the maximum rank of the index;

[0061] Varying the modified normal weight vector W by the Hadamard product of the modified normal weight vector W and the variable weight factor:

[0062]

[0063] Among them, W(X) represents the weight vector of each index after variable weighting; T(X) represents the variable weight factor vector of each index; w j represents the constant weight of the j-th index; T j (x) represents the variable weight factor of the j-th index;

[0064] Then there is:

[0065]

[0066] Among them, w ij represents the weight of the j-th index of the i-th data point; e represents the natural logarithm; V ij represents the rank membership degree of the j-th index of the i-th data point;

[0067] Solving all the preset power quality indexes can obtain the first weight vector.

[0068] The present invention also provides a real-time power quality evaluation system based on cooperative game theory and dynamic entropy value variable weights for the method of the present invention. The system includes a first module, a second module, a third module, and a fourth module;

[0069] The first module is used to collect the time-series data of the preset power quality indexes of the distribution network nodes; perform window sliding normalization on the time-series data to obtain normalized data; and obtain the entropy values of each index and the correlation coefficient matrix between the indexes according to the normalized data;

[0070] The second module is used to obtain the Shapley value of each index according to the entropy value and the correlation coefficient matrix in combination with the cooperative game theory; obtain the constant weight vector of the index according to the Shapley value; calculate the conflict entropy weight of each index, and obtain the expert subjective weight of each index;

[0071] The third module is used to correct the constant weight vector according to the conflict entropy weight and the expert subjective weight to obtain the corrected weight of the index; normalize the corrected weight of the index to obtain the corrected constant weight vector of the index;

[0072] The fourth module is used to perform dynamic variable weighting on the corrected constant weight vector through the variable weight factor under the rank constraint to obtain the first weight vector of the preset power quality indexes; and perform real-time evaluation on the power quality of the distribution network nodes according to the first weight vector.

[0073] The present invention has the following beneficial effects:

[0074] The real-time power quality evaluation method based on cooperative game theory and dynamic entropy value variable weights of the present invention uses a sliding window to perform time-series partitioning on power quality data, avoiding the situation where local anomalies may be weakened due to the broad distribution of overall data in global normalization. The Shapley value of each index is obtained according to the entropy value and the correlation coefficient matrix in combination with the cooperative game theory, and the constant weight vector of the index is obtained according to the Shapley value. Through the cooperative game theory, the alliance analysis of the entropy value is carried out, and the conflict and correlation between the data are organically combined, and the weighting result is more scientific and reasonable; on this basis, the conflict entropy weight of each index is calculated, the expert subjective weight of each index is obtained, and the constant weight vector is corrected according to the conflict entropy weight and the expert subjective weight. The constant weight is corrected by using the dynamic variable weight factor, so that the weighting result can accurately respond to the changes of power quality indexes dynamically, making the method of the present invention have good promotion benefits, universality objectivity, and timeliness accuracy.

[0075] The real-time power quality evaluation system based on cooperative game theory and dynamic entropy value variable weights of the present invention, for the method of the present invention, has the same beneficial effects as the method of the present invention.

[0076] In addition to the purposes, features, and advantages described above, the present invention has other purposes, features, and advantages. The present invention will be further described in detail below with reference to the accompanying drawings. Description of the Drawings

[0077] The accompanying drawings that form a part of the present invention are used to provide a further understanding of the present invention. The schematic embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:

[0078] Figure 1 is a schematic flowchart of the method of the preferred embodiment of the present invention. Detailed Embodiment

[0079] The following will describe the embodiments of the present invention in detail with reference to the accompanying drawings. However, the present invention can be implemented in many different ways defined and covered by the claims.

[0080] See Figure 1 , in the preferred embodiment of the present invention, a real-time power quality evaluation method based on cooperative game theory and dynamic entropy value variable weight is provided, including:

[0081] F1. Collect the time series data of the preset power quality indicators of the distribution network nodes; perform window sliding normalization on the time series data to obtain normalized data; obtain the entropy value of each indicator and the correlation coefficient matrix between the indicators according to the normalized data.

[0082] In the preferred embodiment of the present invention, performing window sliding normalization on the time series data to obtain normalized data includes:

[0083] Divide the window length and the sliding step according to the preset statistical period;

[0084] Confirm the initialization window at the starting point of the time series, move the window by the step length, discard the oldest data point, incorporate the latest data point, and normalize the power quality time series data of each window. The normalization includes:

[0085]

[0086] Among them, represents the normalized value of the s-th data point in the m-th window; x (m) (s) represents the s-th data value in the m-th window; N represents the single window length of the power quality time series data, that is, the total number of sampling points is N; x (m) (k) represents the data value of the k-th data point in the m-th window; k represents the data point in the window, m ≤ k ≤ m + N - 1.

[0087] The above data standardization method erases the differences in dimension and magnitude between different indicators, and the sliding window dynamically adapts to data changes through the standardization of local data, avoiding the distortion caused by global standardization, so as to accurately and stably dynamically evaluate the power quality.

[0088] In the preferred embodiment of the present invention, obtaining the entropy value of each indicator according to the standardized data includes:

[0089] Perform normalization processing on the standardized data:

[0090]

[0091] After performing normalization processing on the standardized data of all M power quality indicators, an initial data matrix X is integrated (m) :

[0092]

[0093] Among them, represents the i-th normalized data of the j-th indicator in the m-th window; represents the standardized value of the first data point in the m-th window; represents the standardized value of the s-th data point in the m-th window; M represents the total number of power quality indicators;

[0094] The entropy value of the indicator includes:

[0095]

[0096] represents the proportion of the i-th sample value of the j-th indicator in the m-th window to the indicator:

[0097]

[0098] When At this time, is meaningless, so is corrected:

[0099]

[0100] Among them, ε is a correction factor. To avoid a huge error in calculating the entropy value, ε takes an extremely small value.

[0101] In the preferred embodiment of the present invention, the correlation coefficient matrix between indicators includes:

[0102] Calculate the Pearson correlation coefficient r between each indicator ij And obtain the correlation coefficient matrix

[0103]

[0104] Among them, r ij represents the Pearson correlation coefficient between index i and index j; x n and y n respectively represent the nth normalized values of index i and index j in the mth time series window; and respectively represent the average normalized data of index i and index j in the mth time series window; represents the correlation coefficient between index i and index j in the mth time series window.

[0105] F2. Obtain the Shapley value of each index according to the entropy value and the correlation coefficient matrix in combination with the cooperative game theory; obtain the constant weight vector of the index according to the Shapley value; calculate the conflict entropy weight of each index, and obtain the subjective weight of each index by experts.

[0106] In the preferred embodiment of the present invention, obtaining the Shapley value of each index according to the entropy value and the correlation coefficient matrix in combination with the cooperative game theory includes:

[0107] Regarding the index as a participant in the cooperative game, define the dynamic income function v(S) of the index coalition S:

[0108]

[0109] Among them, S represents all non-overlapping sets that can be formed by all M indexes, that is, the index coalition, a total of 2 M -1; E j represents the entropy value of the jth index; t is the index of the time window, representing the time step of the dynamic model; τ represents the moment index within the time window; L represents the total window length; α is the time decay factor, and higher weights are given to recent data through the time decay factor, α < 1; x j,τ is the normalized data of index j at the moment τ; β(S) is the sum of the absolute values of the correlation coefficients of all index pairs within the coalition S;

[0110] Then the Shapley value φ j includes:

[0111]

[0112] Among them, A is the set of participants, representing all power quality indexes; A\{j} represents the subset obtained by removing index j from set A; φ j represents the weight Shapley value of the jth index in the entire window; {j} represents the set formed by the jth index; v(S∪{j}) represents the value function of the coalition containing index j.

[0113] In the preferred embodiment of the present invention, the constant weight vector of the index obtained according to the Shapley value includes:

[0114] The constant weight vector of the index includes:

[0115]

[0116] where B j represents the weight of the j-th index in the entire window, and calculating the weight of each index gives the constant weight vector of the index.

[0117] In the preferred embodiment of the present invention, calculating the conflict entropy weight of each index includes:

[0118]

[0119] where Q j is the conflict intensity of index j, P j is the conflict entropy weight, is the average entropy value of the j-th index within the time window; E j is the entropy value of the j-th index; Q c is the conflict intensity of index c; E c is the entropy value of the c-th index.

[0120] F3. Modify the constant weight vector according to the conflict entropy weight and the expert subjective weight to obtain the modified weight of the index; normalize the modified weight of the index to obtain the modified constant weight vector of the index. F3 specifically includes:

[0121] Adopt the idea of cooperative game to modify the constant weight vector with the expert subjective weight and the conflict entropy weight:

[0122] Calculate the consistency correlation coefficient L(h) of the expert subjective weight, the conflict entropy weight and the constant weight vector:

[0123]

[0124] where W j (h) represents the h-th weight of index j, W(d_h) represents the optimal combination weight of the d-1 weights W(1),..., W(h-1), W(h+1),..., W(d) excluding W(h); d represents the total number of weight calculation methods; and respectively represent the average values of W j (h) and W j (d_h);

[0125] Solve for the modified weight W j ' of the index, including:

[0126]

[0127] Normalize the index correction weight W j ' to obtain the corrected normal weight vector W = (w1, w2,..., w M ).

[0128] F4. Dynamically vary the weights of the corrected normal weight vector through the variable weight factors under the rank constraint to obtain the first weight vector of the preset power quality index; perform real-time evaluation of the power quality of the distribution network nodes according to the first weight vector.

[0129] In the preferred embodiment of the present invention, dynamically varying the weights of the corrected normal weight vector through the variable weight factors under the rank constraint to obtain the first weight vector of the preset power quality index includes:

[0130] During the real-time evaluation of power quality, a sudden deterioration of an index with a relatively small constant weight within a very short time variation period will trigger the "barrel effect", that is, in this case, this deteriorating index becomes the "short board" affecting the quality of the entire power quality and should be given a higher weight. However, the constant weights obtained in the above steps can only assign weights to the indices within a certain time window and cannot change the weights at specific time points outside the time window according to the changes of the indices. Considering that the dimensions of each index are different and the degree of index deterioration cannot be intuitively presented by the index data, in order to accurately and dynamically respond to the changes in the weights of each index, the rank limit is used as an index to quantify the degree of index deterioration, and the variable weight factor includes:

[0131] T(x) = e λV(x)

[0132] where T(x) represents the variable weight factor of the index data x at the data point; λ represents the index variable weight factor gain coefficient, and the value of λ affects the variable weight benefit; V(x) is the rank membership function of the index data x at the data point, used to reflect the degree of deterioration of the data x, including:

[0133]

[0134] where x represents the index data of the data point in the time window; V g is the rank coefficient under the g-th rank of the index; [x g-1 , x g is the upper and lower rank limits of the g-th rank; x g represents the highest rank limit under the g-th rank of the index; x g-1 represents the highest rank limit under the (g - 1)-th rank of the index; V gmax represents the maximum rank coefficient of the index; x gmax represents the highest rank limit under the maximum rank of the index;

[0135] Variable weight the modified normal weight vector \(W\) by using the Hadamard product of the modified normal weight vector \(W\) and the variable weight factor:

[0136]

[0137] Among them, \(W(X)\) represents the weight vector of each index after variable weighting; \(T(X)\) represents the variable weight factor vector of each index; \(w\) j represents the constant weight of the \(j\)-th index; \(T\) j (x) represents the variable weight factor of the \(j\)-th index;

[0138] Then there is:

[0139]

[0140] Among them, \(w\) ij represents the weight of the \(j\)-th index of the \(i\)-th data point; \(e\) represents the natural logarithm; \(V\) ij represents the grade membership degree of the \(j\)-th index of the \(i\)-th data point;

[0141] The weight after variable weighting can not only dynamically respond to the impact of the sharp change of the index on the power quality assessment, but also improve the discrimination degree of the index weights in different time domains, making the weighted result more scientifically and accurately reflect the impact of the change of the index data on the comprehensive evaluation of power quality in actual engineering.

[0142] Solve all the preset power quality indexes to obtain the first weight vector.

[0143] The real-time power quality assessment method based on cooperative game theory and dynamic entropy value variable weighting of the present invention uses a sliding window to perform time series partitioning on power quality data, avoiding the situation that local anomalies may be weakened due to the wide distribution of the overall data in global standardization. According to the entropy value and the correlation coefficient matrix, combined with the cooperative game theory, the Shapley value of each index is obtained, and the constant weight vector of the index is obtained according to the Shapley value. Through the cooperative game theory, the alliance analysis of the entropy value is carried out, and the conflict and correlation between the data are organically combined, and the weighted result is more scientific and reasonable; on this basis, the conflict entropy weight of each index is calculated, the expert subjective weight of each index is obtained, and the normal weight vector is modified according to the conflict entropy weight and the expert subjective weight. The constant weight is modified by using the dynamic variable weight factor, so that the weighted result can accurately and dynamically respond to the change of the power quality index, making the method of the present invention have good promotion benefits, universality objectivity and timeliness accuracy.

[0144] In a preferred embodiment of the present invention, a real-time power quality assessment system based on cooperative game theory and dynamic entropy value variable weighting is also provided for the method of the present invention. The system includes a first module, a second module, a third module and a fourth module;

[0145] The first module is used to collect the time - series data of the preset power quality indicators of the distribution network nodes; perform window - sliding standardization on the time - series data to obtain standardized data; obtain the entropy values of each indicator and the correlation coefficient matrix between indicators based on the standardized data;

[0146] The second module is used to obtain the Shapley values of each indicator according to the entropy values and the correlation coefficient matrix in combination with the cooperative game theory; obtain the constant weight vector of the indicators according to the Shapley values; calculate the conflict entropy weights of each indicator, and obtain the subjective weights of experts for each indicator;

[0147] The third module is used to correct the constant weight vector according to the conflict entropy weights and the subjective weights of experts to obtain the corrected weight of the indicator; normalize the corrected weight of the indicator to obtain the corrected constant weight vector of the indicator;

[0148] The fourth module is used to perform dynamic weighting on the corrected constant weight vector through the variable weight factor under the rank constraint to obtain the first weight vector of the preset power quality indicators; perform real - time evaluation of the power quality of the distribution network nodes according to the first weight vector.

[0149] The real - time power quality evaluation system based on the cooperative game theory and dynamic entropy - value variable weighting of the present invention is used for the method of the present invention and has the same beneficial effects as the method of the present invention.

[0150] Verification part:

[0151] To verify the effectiveness of the present invention, first, use a power quality monitoring platform of a certain distribution network to obtain the historical power quality monitoring data of the monitoring points of the 380V voltage level, and extract the voltage deviation, frequency deviation, three - phase unbalance degree, harmonic distortion rate, and voltage fluctuation and flicker data within a certain day. According to the periodic curve of the electrical load, set the sliding window length to 10 minutes, the sliding step to 1 minute, each window contains 20 consecutive sampling points, and take the data of the m - th window for weight example calculation. The original data is shown in Table 1.

[0152] Calculate the data standardization matrix according to the original data, as shown in Table 2. Then calculate the information entropy of the indicators and the Pearson correlation coefficient between the indicators. The entropy values of the indicators are shown in Table 3. The Pearson correlation coefficient matrix between the indicators is shown in Table 4.

[0153] Calculate the dynamic benefits of each coalition for power quality evaluation according to all the coalitions that can be formed by the indicators. On this basis, calculate the Shapley values of each indicator, and thus obtain the constant weight of each indicator.

[0154] Subsequently, calculate the conflict entropy weights of each indicator, regard the subjective weight and the conflict entropy weight as a whole and form an alliance, with the subjective weight of the expert A j and the conflict entropy weight P jTo correct the constant weights of the Shapley value, each constant weight value is shown in Table 5.

[0155] According to the national relevant standards of power quality, divide the limit interval as shown in Table 6, calculate the variable weight factors, and finally use the Hadamard product of the constant weight vector and the variable weight factors to vary the constant weight vector W to obtain the dynamic evaluation value. Part of the variable weight results are shown in Table 7. As can be seen from the above, a higher weight is given when a certain index deteriorates sharply.

[0156] Table 1 Original data table of power quality indicators within the m-th window

[0157]

[0158] Table 2 Data standardization data table

[0159]

[0160]

[0161] Table 3 Entropy value table of indicators

[0162]

[0163] Table 4 Pearson correlation coefficient matrix between indicators

[0164]

[0165] Table 5 Constant weight of each indicator

[0166]

[0167] Table 6 Classification of power quality indicator levels

[0168]

[0169] Table 7 Dynamic weights of power quality indicators at each time point

[0170]

[0171]

[0172] The above are only the preferred embodiments of the present invention and are not used to limit the present invention. For those skilled in the art, the present invention can have various changes and modifications. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A real-time power quality assessment method based on cooperative game theory and dynamic entropy variable weight, characterized in that, Including: Collecting the time - series data of the preset power quality indicators of the distribution network nodes; Performing window - sliding normalization on the time - series data to obtain normalized data; Obtaining the entropy values of each indicator and the correlation coefficient matrix between indicators according to the normalized data; Obtaining the Shapley values of each indicator by combining the entropy values and the correlation coefficient matrix with the cooperative game theory; obtaining the constant weight vector of the indicators according to the Shapley values; calculating the conflict entropy weights of each indicator, and obtaining the expert subjective weights of each indicator; Correcting the constant weight vector according to the conflict entropy weights and the expert subjective weights to obtain the corrected weight vector of the indicators; Normalizing the corrected weight vector of the indicators to obtain the corrected normal weight vector of the indicators; Performing dynamic weight variation on the corrected normal weight vector through the variable weight factor under hierarchical constraints to obtain the first weight vector of the preset power quality indicators; performing real - time assessment of the power quality of the distribution network nodes according to the first weight vector.

2. The real-time power quality assessment method based on cooperative game theory and dynamic entropy value variable weight according to claim 1, characterized in that Performing window - sliding normalization on the time - series data to obtain normalized data includes: Dividing the window length and the sliding step according to the preset statistical period; Confirming the initialization window at the starting point of the time series, moving the window by the step, discarding the oldest data point, incorporating the latest data point, and normalizing the power quality time - series data of each window, and the normalization includes: Among them, represents the normalized value of the s-th data point in the m-th window; x (m) (s) represents the s-th data value in the m-th window; N represents the length of a single window of power quality time series data, that is, the total number of sampling points is N; x (m) (k) represents the data value of the k-th data point in the m-th window; k represents the data point in the window, m ≤ k ≤ m + N - 1.

3. The real-time power quality evaluation method based on cooperative game theory and dynamic entropy value variable weight according to claim 2, characterized in that Obtaining the entropy values of each indicator according to the normalized data includes: Performing normalization processing on the normalized data: After normalizing the standardized data of all M power quality indicators, the data initialization matrix X is integrated and obtained. (m) : Among them, represents the i-th normalized data of the j-th index in the m-th window; represents the standardized value of the first data point in the m-th window; represents the standardized value of the s-th data point in the m-th window; M represents the total number of power quality indicators. The entropy value of the indicator includes: Indicates the proportion of the i-th sample value of the j-th indicator in the m-th window to this indicator: When , is meaningless, so correct : Where ε is a correction factor. To avoid a huge error in calculating the entropy value, ε takes an extremely small value.

4. The real-time power quality evaluation method based on cooperative game theory and dynamic entropy value variable weight according to claim 3, characterized in that The correlation coefficient matrix between the indicators includes: Calculate the Pearson correlation coefficient r between each indicator ij And obtain the correlation coefficient matrix Among them, r ij represents the Pearson correlation coefficient between index i and index j; x n and y n respectively represent the nth normalized values of index i and index j in the mth time series window; and respectively represent the average normalized data of index i and index j in the mth time series window; represents the correlation coefficient between index i and index j in the mth time series window.

5. The real-time power quality assessment method based on cooperative game theory and dynamic entropy value variable weight according to claim 4, characterized in that Obtaining the Shapley values of each indicator by combining the entropy values and the correlation coefficient matrix with the cooperative game theory includes: Regarding the indicators as participants in the cooperative game, defining the dynamic revenue function v(S) of the indicator coalition S: Among them, S represents all non-overlapping sets that can be formed by all M indicators, that is, the indicator coalition, with a total of 2 M - 1; E j represents the entropy value of the j-th indicator; t is the index of the time window, representing the time step of the dynamic model; τ represents the moment index within the time window; L represents the total window length; α is the time decay factor, which assigns higher weights to recent data through the time decay factor, α < 1; x j,τ is the normalized data of indicator j at moment τ; β(S) is the sum of the absolute values of the correlation coefficients of all indicator pairs within coalition S; Then the Shapley value φ of the metric j includes: Among them, \(A\) is the set of participants, representing all power quality indicators; \(A\setminus\{j\}\) represents the subset of set \(A\) after removing indicator \(j\); \(\varphi\) j represents the Shapley value of the weight of the \(j\)-th indicator in the entire window; \(\{j\}\) represents the set formed by the \(j\)-th indicator; \(v(S\cup\{j\})\) represents the value function of the coalition including indicator \(j\).

6. The real-time power quality assessment method based on cooperative game theory and dynamic entropy value variable weight according to claim 5, characterized in that Obtaining the constant weight vector of the indicators according to the Shapley values includes: The constant weight vector of the indicator includes: Among them, B j represents the weight of the j-th index in the entire window. Calculating the weight of each index gives the constant weight vector of the said index.

7. The real-time power quality assessment method based on cooperative game theory and dynamic entropy value variable weight according to claim 6, characterized in that Calculating the conflict entropy weights of each indicator includes: Among them, Q j is the conflict intensity of index j, and P j is the conflict entropy weight, is the average entropy value of the j-th index within the time window; E j is the entropy value of the j-th index; Q c is the conflict intensity of index c; E c is the entropy value of the c-th index.

8. The real-time power quality assessment method based on cooperative game theory and dynamic entropy weight variable weight according to claim 7, characterized in that Correcting the constant weight vector according to the conflict entropy weights and the expert subjective weights to obtain the corrected weight vector of the indicators; Normalizing the corrected weight vector of the indicators to obtain the corrected normal weight vector of the indicators includes: Adopting the idea of cooperative game to correct the constant weight vector with the expert subjective weights and the conflict entropy weights: Calculating the consistency correlation coefficient L(h) of the expert subjective weights, the conflict entropy weights and the constant weight vector: Among them, W j (h) represents the h-th weight of index j, and W(d_h) represents the optimal combined weight of the d - 1 weights W(1),..., W(h - 1), W(h + 1),..., W(d) excluding W(h); d represents the total number of weight calculation methods; and respectively represent the average values of W j (h) and W j (d_h); Solve for the corrected weight \(W\) of the said index j ', including: Normalize the corrected weight W of the said index j ' to obtain the corrected normal weight vector W=(w1, w2,..., w M ).

9. The real-time power quality assessment method based on cooperative game theory and dynamic entropy value variable weight according to claim 8, characterized in that, Performing dynamic weight variation on the corrected normal weight vector through the variable weight factor under hierarchical constraints to obtain the first weight vector of the preset power quality indicators includes: Taking the hierarchical limit value as an indicator to quantify the degree of deterioration of the indicator, then the variable weight factor includes: T(x) = e λV(x) Where T(x) represents the variable weight factor of the indicator data x of the data point; λ represents the variable weight factor gain coefficient, and the value of λ affects the variable weight benefit; V(x) is the hierarchical membership function of the indicator data x of the data point, used to reflect the degree of deterioration of the data x, including: Among them, x represents the index data of the data points in the time window; V g is the grade coefficient at the g-th grade of the index; [x g-1 , x g ) are the upper and lower grade limits of the g-th grade; x g represents the highest grade limit at the g-th grade of the index; x g-1 represents the highest grade limit at the (g - 1)-th grade of the index; V gmax represents the maximum grade coefficient of the index; x gmax represents the highest grade limit at the maximum grade of the index; Performing weight variation on the corrected normal weight vector W by using the Hadamard product of the corrected normal weight vector W and the variable weight factor: Among them, W(X) represents the weight vector of each index after variable weight; T(X) represents the variable weight factor vector of each index; w j represents the constant weight of the j-th index; T j (x) represents the variable weight factor of the j-th index; Then there is: where, w ij represents the weight of the j-th index of the i-th data point; e represents the natural logarithm; V ij represents the grade membership of the j-th index of the i-th data point; Solving for all the preset power quality indicators, the first weight vector can be obtained.

10. A real-time power quality evaluation system based on cooperative game theory and dynamic entropy value variable weights, for the method according to any one of claims 1 to 9, characterized in that, The system includes a first module, a second module, a third module, and a fourth module; The first module is used to collect the time-series data of the preset power quality indexes of the distribution network nodes; perform window sliding normalization on the time-series data to obtain normalized data; obtain the entropy values of each index and the correlation coefficient matrix between the indexes according to the normalized data; The second module is used to obtain the Shapley value of each index according to the entropy value and the correlation coefficient matrix in combination with the cooperative game theory; obtain the constant weight vector of the indexes according to the Shapley value; calculate the conflict entropy weight of each index, and obtain the expert subjective weight of each index; The third module is used to correct the constant weight vector according to the conflict entropy weight and the expert subjective weight to obtain the corrected weight of the index; Normalize the corrected weight of the index to obtain the corrected constant weight vector of the index; The fourth module is used to perform dynamic weighting on the corrected constant weight vector through the variable weight factor under the rank constraint to obtain the first weight vector of the preset power quality index; perform real-time evaluation on the power quality of the distribution network nodes according to the first weight vector.

Images