Partitioning method and device for power quality management of power distribution network
By calculating node voltage sensitivity and grey relational degree, and combining spectral clustering algorithm, the problem of inaccurate node zoning in power quality management of distribution networks is solved, enabling more reasonable area division and equipment installation, and improving the power quality improvement effect.
Patent Information
- Application Number
- CN202411138125.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-19
- Publication Date
- 2024-11-22
- Estimated Expiration
- 2044-08-19
AI Technical Summary
In the current technology for power quality management of distribution networks, the node zoning method cannot accurately reflect the actual correlation of nodes, resulting in slow area division and low correlation, which cannot effectively guide the installation of management equipment.
By calculating the reactive voltage sensitivity and active voltage sensitivity of nodes, and combining grey relational analysis and spectral clustering algorithms, a comprehensive correlation matrix is constructed to determine the electrical distance, trend of change, and amplitude correlation between nodes. The spectral clustering algorithm is used to divide the region, and the dominant node is selected to install the treatment equipment.
It has enabled more accurate zoning, improved the effectiveness of power quality improvement, reduced the installation cost of monitoring equipment, and improved the installation accuracy of treatment equipment.
Smart Images

Figure CN119009997A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of power distribution network management, and in particular relates to a power quality management zoning method and device for a power distribution network. Background Art
[0002] Nowadays, with the access of a large number of power electronic devices such as electric vehicles, air conditioners, and distributed photovoltaics to the distribution network, the distribution of power quality pollution sources in the distribution network is showing a trend of randomization, decentralization, and full-network. The traditional method of point-to-point governance of pollution from the main power quality disturbance sources can no longer meet the power quality needs of the entire distribution network, and a set of power quality governance strategies that can be coordinated from the whole network level are urgently needed. Among them, the current hot research direction is to divide the distribution network into zones and carry out power governance for regional leading nodes, and to use the point-to-surface approach.
[0003] The prior art proposes a strategy for power quality zoning based on the electrical distance of nodes. Nodes with close electrical distances are divided into one area. However, zoning based solely on electrical distance cannot accurately reflect the actual power quality changes of nodes in the area. The correlation between different nodes is not only related to the electrical distance but also to many factors such as the distribution of disturbance sources. Zoning based solely on electrical distance cannot reflect the true degree of correlation between nodes.
[0004] The prior art proposes a power quality zoning management strategy based on the historical power quality measurement data of the existing power grid. The node correlation is calculated according to the historical power quality change trends of different nodes. It is believed that the closer the historical power quality change trends of the nodes are, the higher the correlation between the two nodes. However, this method simply relies on the power quality change trends of the nodes, ignoring the randomness of the change trends between nodes. It is possible that the change trends of two nodes in the region are very similar just by chance, but in fact the electrical distance between the two nodes is very far and the correlation is very low. This will cause two nodes with very low actual correlation to be divided into the same area, which is obviously not in line with the actual situation. Summary of the invention
[0005] In view of the above-mentioned deficiencies in the prior art, the present invention provides a distribution network power quality management zoning method and device, which solves the problems of slow zone division speed and low actual correlation of nodes in the divided areas.
[0006] In order to achieve the above objectives, the technical solution adopted by the present invention is as follows: On the one hand, the present invention provides a distribution network power quality management zoning method, comprising the following steps:
[0007] S1. According to the network topology and parameters of the distribution network, the reactive voltage sensitivity and active voltage sensitivity of each node in the distribution network are calculated using the power flow equation, and the electrical distance between each node is obtained according to the reactive voltage sensitivity and active voltage sensitivity of each node;
[0008] S2, constructing a time series measurement data set using historical measurement data of harmonics, three-phase unbalance and voltage deviation at each node;
[0009] S3, using a normalization method to normalize the time series measurement data set, and using grey correlation to analyze the normalized time series measurement data set, to obtain the change trend correlation and amplitude correlation of each node respectively;
[0010] S4, determining the comprehensive correlation between each node according to the correlation of the change trend, the amplitude correlation and the electrical distance of each node, and integrating the comprehensive correlation between each node to obtain a comprehensive correlation matrix;
[0011] S5. Use the spectral clustering algorithm to cluster the comprehensive correlation matrix and combine it with the elbow method to determine the final number of regional divisions. According to the final number of regional divisions, regional division for harmonic, three-phase unbalance and voltage deviation management is achieved;
[0012] S6. According to the regional division results, select the node with the largest sum of comprehensive correlations with all nodes in any area as the dominant node in the area, and install management equipment on the dominant node to complete the power quality management zoning of the distribution network.
[0013] The beneficial effects of the present invention are as follows: the present invention evaluates the correlation between different nodes from multiple dimensions such as electrical distance, change trend and amplitude, and is more comprehensive than existing partitioning technologies. The regional division results are more reasonable and will not be affected by chance, avoiding the unreasonable phenomenon of nodes with low correlation being divided into the same area. The dominant nodes in the area can be accurately identified to guide the installation of relevant governance equipment and improve the overall power quality improvement level in the area. It can also be used to guide the installation of monitoring devices, and use the dominant nodes to represent the overall regional change level, thereby reducing the installation cost of distribution network monitoring equipment.
[0014] Furthermore, the S1 comprises the following steps:
[0015] S101. Calculate reactive voltage sensitivity and active voltage sensitivity of each node based on the Newton-power flow equation according to the network topology and parameters of the distribution network;
[0016] S102, obtaining the voltage sensitivity between the nodes using a sensitivity equation according to the reactive voltage sensitivity and the active voltage sensitivity of each node;
[0017] S103 . Calculate the electrical distance between the nodes using an electrical distance formula according to the voltage sensitivity between the nodes.
[0018] Furthermore, the equation for the voltage sensitivity is as follows:
[0019]
[0020] Among them, A ij represents the voltage sensitivity between node i and node j, Represents the voltage amplitude of node j V j Active voltage sensitivity at node i, Represents the voltage amplitude of node i V i Active voltage sensitivity at node j, Represents the voltage amplitude of node j V j Reactive voltage sensitivity to node i, Represents the voltage amplitude of node i V i The reactive voltage sensitivity of node j is: Represents the voltage amplitude of node i V i The partial derivative of ΔP i represents the amount of change in active power at node i, ΔQ i Represents the change in reactive power at node i.
[0021] Furthermore, the S3 comprises the following steps:
[0022] S301, normalizing the historical measurement data of harmonics, three-phase unbalance and voltage deviation of each node according to the time series measurement data set, and mapping the historical measurement data to the [0,1] interval;
[0023] S302, using the grey correlation analysis method to analyze the correlation of the changing trends of harmonics, three-phase imbalance and voltage deviation at different nodes according to the normalized time series measurement data set;
[0024] S303 , according to the normalized time series measurement data set, the grey correlation analysis method is used to analyze the amplitude correlation of harmonics, three-phase unbalance and voltage deviation of different nodes.
[0025] The beneficial effect of the above further scheme is: the present invention adopts the grey correlation analysis method to accurately analyze the similarity correlation of indicators of different dimensions of different nodes, and provide a reasonable basis for subsequent regional division.
[0026] Furthermore, the S5 comprises the following steps:
[0027] S501, using a spectral clustering algorithm to divide the distribution network into regions, calculating a degree matrix based on a comprehensive correlation matrix of harmonics, three-phase imbalance, and voltage deviation at each node, and symmetric normalizing the degree matrix to obtain a Laplace matrix;
[0028] S502, according to the eigenvalues and eigenvectors of the Laplace matrix, selecting the eigenvectors corresponding to the first k smallest eigenvalues to form an eigenvector matrix;
[0029] S503, taking each row of the eigenvector matrix as a data point, and clustering the data points using the k-means algorithm in the eigenspace;
[0030] S504, according to the clustering results, using the elbow method to determine the number of clusters, analyzing the value changes in the k-means algorithm under different cluster numbers, and determining the inflection point position where the value in the k-means algorithm drops sharply as k increases;
[0031] S505 . According to the inflection point position, the number of clusters corresponding to the inflection point position is selected as the optimal number of clusters, and the optimal number of clusters is used as the final number of region divisions.
[0032] Furthermore, the degree matrix F is a diagonal matrix, and the diagonal elements F ii is the sum of the comprehensive associations between all nodes connected to node i, and the diagonal element F ii The formula is as follows:
[0033]
[0034] Among them, Z ij Represents the element in the i-th row and j-th column of the comprehensive correlation matrix;
[0035] The formula of the Laplace matrix L is as follows:
[0036] L=IF 1 / 2 ZF -1 / 2
[0037] Among them, I represents the identity matrix and Z represents the comprehensive correlation matrix.
[0038] The beneficial effects of the above further scheme are as follows: the present invention adopts the spectral clustering algorithm and the elbow method to achieve regional division that does not depend on the initial point, does not need to pre-set the number of partitions, and more objectively determines the number of partitions based on actual correlation data. It can also perform effective clustering in high-dimensional data, reducing the complexity and noise interference of data in high-dimensional space.
[0039] Furthermore, the calculation formula of the comprehensive correlation is as follows:
[0040]
[0041] Among them, ε represents the comprehensive correlation, Indicates the correlation degree of harmonic amplitude of each node, Indicates the correlation degree of harmonic change trend of each node, D hIndicates the electrical distance, b 1 、b 2 、b 3 Both represent the weight coefficients of amplitude correlation, change trend correlation and electrical distance, and b 1 +b 2 +b 3 =1;
[0042] The comprehensive correlation matrix is as follows:
[0043]
[0044] Among them, ε ij Represents the comprehensive correlation between the i-th node and the j-th node.
[0045] On the other hand, the present invention provides a distribution network power quality management zoning device, characterized in that it includes:
[0046] Obtaining electrical distance module: used to calculate the reactive voltage sensitivity and active voltage sensitivity of each node in the distribution network using the power flow equation according to the network topology and parameters of the distribution network, and to obtain the electrical distance between each node according to the reactive voltage sensitivity and active voltage sensitivity of each node;
[0047] Dataset building module: used to build a time series measurement dataset using the historical measurement data of harmonics, three-phase unbalance and voltage deviation of each node;
[0048] Correlation analysis module: used to normalize the time series measurement data set using the normalization method, and use the grey correlation to analyze the normalized time series measurement data set, and obtain the change trend correlation and amplitude correlation of each node respectively;
[0049] Comprehensive correlation module: used to determine the comprehensive correlation between nodes according to the correlation of the change trend, amplitude and electrical distance of each node, and integrate the comprehensive correlation between nodes to obtain the comprehensive correlation matrix;
[0050] Spectral clustering module: used to cluster the comprehensive correlation matrix using the spectral clustering algorithm and determine the final number of regional divisions in combination with the elbow method. According to the final number of regional divisions, regional division for harmonic, three-phase unbalance and voltage deviation management is achieved;
[0051] Selection module: It is used to select the node with the largest sum of comprehensive correlation with all nodes in any area as the dominant node in the area according to the regional division results, and install management equipment on the dominant node to complete the power quality management zoning of the distribution network.
[0052] Furthermore, the module for obtaining the electrical distance includes:
[0053] Newton-power flow equation submodule: used to calculate the reactive voltage sensitivity and active voltage sensitivity of each node based on the Newton-power flow equation according to the network topology and parameters of the distribution network;
[0054] Voltage sensitivity submodule: used to obtain the voltage sensitivity between each node based on the reactive voltage sensitivity and active voltage sensitivity of each node using the sensitivity equation;
[0055] Electrical distance submodule: used to calculate the electrical distance between each node based on the voltage sensitivity between each node using the electrical distance formula.
[0056] Furthermore, the association analysis module includes:
[0057] Normalization submodule: used to normalize the historical measurement data of harmonics, three-phase unbalance and voltage deviation of each node according to the time series measurement data set, and map the historical measurement data to the [0,1] interval;
[0058] Change trend correlation submodule: It is used to analyze the change trend correlation of harmonics, three-phase unbalance and voltage deviation of different nodes using grey correlation analysis method according to the normalized time series measurement data set;
[0059] Amplitude correlation submodule: It is used to analyze the amplitude correlation of harmonics, three-phase unbalance and voltage deviation at different nodes using grey correlation analysis method according to the normalized time series measurement data set.
[0060] Furthermore, the spectral clustering module includes:
[0061] Degree matrix submodule: used to calculate the degree matrix according to the comprehensive correlation matrix of harmonics, three-phase unbalance and voltage deviation of each node, and symmetric normalize the degree matrix to obtain the Laplace matrix;
[0062] Eigenvector submodule: used to select the eigenvectors corresponding to the first k smallest eigenvalues according to the eigenvalues and eigenvectors of the Laplace matrix to form an eigenvector matrix;
[0063] K-means submodule: It is used to treat each row of the feature vector matrix as a data point and cluster the data points using the K-means algorithm in the feature space;
[0064] Elbow method submodule: used to determine the number of clusters based on the clustering results using the elbow method, analyze the value changes in the k-means algorithm under different cluster numbers, and determine the inflection point where the value in the k-means algorithm drops sharply as k increases;
[0065] Optimal number of clusters submodule: It is used to select the number of clusters corresponding to the inflection point position as the optimal number of clusters according to the inflection point position, and use the optimal number of clusters as the final number of region divisions.
[0066] The present invention provides a computer device, characterized in that it comprises: one or more processors;
[0067] The processor is used to store one or more programs; when the one or more programs are executed by the one or more processors, the distribution network power quality management zoning method as described in any one of the above items is implemented.
[0068] The present invention provides a computer-readable storage medium, characterized in that a computer program is stored thereon, and when the computer program is executed, the distribution network power quality management zoning method as described in any one of the above items is implemented.
[0069] The beneficial effects of the above further scheme are: the present invention provides a distribution network power quality management zoning device, which evaluates the correlation between different nodes from multiple dimensions such as electrical distance, change trend and amplitude. Compared with the existing zoning technology, it is more comprehensive and the regional division results are more reasonable, which reduces the installation cost of distribution network monitoring equipment. BRIEF DESCRIPTION OF THE DRAWINGS
[0070] Figure 1 The figure is a flow chart of the method of the present invention.
[0071] Figure 2 This is the IEEE33-node distribution network diagram.
[0072] Figure 3 This is a diagram showing the result of dividing the 5th harmonic content rate using the method of the present invention.
[0073] Figure 4 This is a diagram showing the classification results of the 5th harmonic content rate in the prior art.
[0074] Figure 5 This is a comparison chart before and after installing the harmonic control device at the harmonic-dominant node.
[0075] Figure 6 This is a comparison diagram before and after installing the three-phase unbalance control device at the three-phase unbalance dominant node.
[0076] Figure 7 This is a comparison chart before and after installing the voltage management device at the node with dominant voltage deviation.
[0077] Figure 8 It is a schematic diagram of the device structure of the present invention. DETAILED DESCRIPTION
[0078] The specific implementation modes of the present invention are described below so that those skilled in the art can understand the present invention. However, it should be clear that the present invention is not limited to the scope of the specific implementation modes. For those of ordinary skill in the art, as long as various changes are within the spirit and scope of the present invention as defined and determined by the attached claims, these changes are obvious, and all inventions and creations utilizing the concept of the present invention are protected.
[0079] Example 1
[0080] like Figure 1 As shown, the present invention provides a distribution network power quality management zoning method, and its implementation method is as follows:
[0081] S1. According to the network topology and parameters of the distribution network, the reactive voltage sensitivity and active voltage sensitivity of each node in the distribution network are calculated using the power flow equation, and the electrical distance between each node is obtained according to the reactive voltage sensitivity and active voltage sensitivity of each node. The implementation method is as follows:
[0082] S101. Calculate reactive voltage sensitivity and active voltage sensitivity of each node based on the Newton-power flow equation according to the network topology and parameters of the distribution network;
[0083] S102, obtaining the voltage sensitivity between the nodes using a sensitivity equation according to the reactive voltage sensitivity and the active voltage sensitivity of each node;
[0084] S103 . Calculate the electrical distance between the nodes using an electrical distance formula according to the voltage sensitivity between the nodes.
[0085] In this embodiment, Figure 2 As shown, according to the network topology and parameters, the reactive voltage sensitivity and active voltage sensitivity of each node are calculated based on the power flow equation, and the electrical distance d1 between each node is further obtained.
[0086] In this embodiment, the electrical distance is the most obvious indicator of the correlation between two nodes. The closer the electrical distance between the two nodes, the greater the mutual influence between them. The electrical distance can be calculated according to the Newton-power flow equation in polar coordinates. The formula of the Newton-power flow equation is as follows:
[0087]
[0088] Among them, ΔP represents the transformation vector of node active power, ΔQ represents the transformation vector of node reactive power, Δθ represents the transformation vector of node phase angle, ΔV represents the transformation vector of node voltage, H, N, K and L represent different block matrices, and the specific expressions are as follows:
[0089]
[0090] Among them, H ij represents the element in the i-th row and j-th column of the block matrix H, N ij represents the element in the i-th row and j-th column of the block matrix N, K ij represents the element in the i-th row and j-th column of the block matrix K, L ij represents the element in the i-th row and j-th column of the block matrix L, V j represents the voltage at node j, Represents the active power P of node i i The change in phase angle θ at node j j The partial derivative of Represents the active power P of node i i The change in the voltage amplitude V at node j is j The partial derivative of represents the reactive power Q of node i i The change in phase angle θ at node j j The partial derivative of Represents the reactive power Q of node i i The change in the voltage amplitude V at node j is j The partial derivative of .
[0091] Therefore, according to N ij With L ij The node voltage for the active power P of node i can be obtained i and reactive power Q of node i i Sensitivity and Further, according to the voltage sensitivity formula, the voltage sensitivity A between nodes can be obtained: ij , the voltage sensitivity equation is shown below:
[0092]
[0093] Among them, A ij represents the voltage sensitivity between node i and node j, Represents the voltage amplitude of node j V j Active voltage sensitivity at node i, Represents the voltage amplitude of node i V i Active voltage sensitivity at node j, Represents the voltage amplitude of node j V j Reactive voltage sensitivity to node i, Represents the voltage amplitude of node i V i The reactive voltage sensitivity of node j is: Represents the voltage amplitude of node i V i The partial derivative of ΔP irepresents the amount of change in active power at node i, ΔQ i Represents the transformation amount of reactive power of node i;
[0094] According to the voltage sensitivity A ij The electrical distance D between nodes can be further obtained ij , the electrical distances are as follows:
[0095] D ij =log(A ij *A ji )
[0096] Wherein, log(·) represents the logarithmic function.
[0097] S2, constructing a time series measurement data set using historical measurement data of harmonics, three-phase unbalance and voltage deviation at each node;
[0098] In this embodiment, historical measurement data of harmonics, three-phase imbalance and voltage deviation of each node are obtained.
[0099] S3. Use the normalization method to normalize the time series measurement data set, and use the grey correlation degree to analyze the normalized time series measurement data set to obtain the change trend correlation degree and amplitude correlation degree of each node. The implementation method is as follows:
[0100] S301, normalizing the historical measurement data of harmonics, three-phase unbalance and voltage deviation of each node according to the time series measurement data set, and mapping the historical measurement data to the [0,1] interval;
[0101] S302, using the grey correlation analysis method to analyze the correlation of the changing trends of harmonics, three-phase imbalance and voltage deviation at different nodes according to the normalized time series measurement data set;
[0102] S303 , according to the normalized time series measurement data set, the grey correlation analysis method is used to analyze the amplitude correlation of harmonics, three-phase unbalance and voltage deviation of different nodes.
[0103] In this embodiment, the historical measurement data of harmonics, three-phase imbalance and voltage deviation are normalized, and grey correlation analysis is used to obtain the similarities dH1, dI1 and dV1 of the harmonics, three-phase imbalance and voltage deviation change trends of each node.
[0104] In this embodiment, based on the historical measurement data of harmonics, three-phase imbalance and voltage deviation, the similarities dH2, dI2 and dV2 between different power quality amplitudes are analyzed using grey correlation.
[0105] In this embodiment, when calculating the changing trends of harmonics, three-phase imbalance and voltage deviation at different nodes, it is necessary to normalize the measured data and map them to the [0,1] interval. The normalization formula is as follows:
[0106]
[0107] Where X represents the j-node harmonic, three-phase unbalance and voltage deviation time series measurement data set, x j (k) They represent the harmonic, three-phase unbalance and voltage deviation measurement data of node j at the kth moment before and after normalization;
[0108] In this embodiment, after obtaining the harmonic, three-phase unbalance and voltage deviation measurement data set X of different nodes and the normalized data set Y, the gray correlation analysis method needs to be used to analyze the amplitude similarity and transformation trend similarity of different nodes. Taking the harmonic amplitude data set X of different nodes as an example, the correlation of different node amplitudes is analyzed. First, the correlation coefficient between node k and other nodes is calculated. The formula of the correlation coefficient is as follows:
[0109]
[0110] Among them, δ k (t) represents the grey relational coefficient of the kth comparison sequence at the tth moment, x 0 (t) represents the reference sequence, x k (t) represents the comparison sequence, ρ represents the resolution coefficient, which is usually set to 0.5, min k min t |x 0 (t)-x k (t)| represents the minimum difference between the two poles, max k max t |x 0 (t)-x k (t)| represents the maximum difference between the two poles;
[0111] After obtaining the grey relational coefficient, calculate the grey relational degree λ k , the grey relational degree λ k The formula is as follows:
[0112]
[0113] Where n represents the number of nodes;
[0114] The formula of the correlation matrix T is as follows:
[0115]
[0116] Among them, λij Represents the grey relational degree between the i-th node and the j-th node.
[0117] S4, determining the comprehensive correlation between each node according to the correlation of the change trend, the amplitude correlation and the electrical distance of each node, and integrating the comprehensive correlation between each node to obtain a comprehensive correlation matrix;
[0118] In this embodiment, different weights are assigned to the similarity indicators, and the comprehensive similarity indicators dH3, dI3, dV3 of the harmonics, three-phase imbalance and voltage deviation between the nodes are obtained by adding them.
[0119] In this embodiment, since the three types of power quality, harmonics, three-phase imbalance and voltage deviation, have already obtained their amplitude correlation matrix, change trend correlation matrix and electrical matrix respectively, it is necessary to establish a comprehensive similarity index to evaluate the correlation of each node. Taking harmonics as an example, the formula for its comprehensive correlation is as follows:
[0120]
[0121] Among them, ε represents the comprehensive correlation, Indicates the correlation degree of harmonic amplitude of each node, Indicates the correlation degree of harmonic change trend of each node, D h Indicates the electrical distance, b 1 、b 2 、b 3 represents different weight coefficients, and b 1 +b 2 +b 3 = 1. The comprehensive correlation matrix Z can be obtained by calculating the comprehensive correlation between each node and other nodes.
[0122] The comprehensive correlation matrix is as follows:
[0123]
[0124] Among them, ε ij Represents the comprehensive correlation between the i-th node and the j-th node.
[0125] S5. Use the spectral clustering algorithm to cluster the comprehensive correlation matrix and combine it with the elbow method to determine the final number of regional divisions. According to the final number of regional divisions, regional division for harmonic, three-phase unbalance and voltage deviation management is achieved. The implementation method is as follows:
[0126] S501, using a spectral clustering algorithm to divide the distribution network into regions, calculating a degree matrix based on a comprehensive correlation matrix of harmonics, three-phase imbalance, and voltage deviation at each node, and symmetric normalizing the degree matrix to obtain a Laplace matrix;
[0127] S502, according to the eigenvalues and eigenvectors of the Laplace matrix, selecting the eigenvectors corresponding to the first k smallest eigenvalues to form an eigenvector matrix;
[0128] S503, taking each row of the eigenvector matrix as a data point, and clustering the data points using the k-means algorithm in the eigenspace;
[0129] S504, according to the clustering results, using the elbow method to determine the number of clusters, analyzing the value changes in the k-means algorithm under different cluster numbers, and determining the inflection point position where the value in the k-means algorithm drops sharply as k increases;
[0130] S505 . According to the inflection point position, the number of clusters corresponding to the inflection point position is selected as the optimal number of clusters, and the optimal number of clusters is used as the final number of region divisions.
[0131] In this embodiment, a spectral clustering algorithm is used to set corresponding thresholds according to the comprehensive similarity indicators dH3, dI3, and dV3 of different nodes to divide the harmonic, three-phase unbalance, and voltage deviation management areas.
[0132] In this embodiment, the comprehensive correlation matrix Z reflects the degree of coupling between different nodes. The higher the correlation, the closer the relationship between the two nodes. Governing one of the nodes will have a better effect on improving other nodes, and the power quality of the node can better represent the power quality indicators of other nodes. Therefore, the clustering algorithm can be used for partitioning, and the spectral clustering algorithm can be selected for regional division, as follows:
[0133] First, the degree matrix F is calculated based on the correlation matrix Z. F is a diagonal matrix. The diagonal elements F ii is the sum of the comprehensive associations between all nodes connected to node i, and the diagonal element F ii The formula is as follows:
[0134]
[0135] Then the degree matrix is symmetrically normalized to obtain the Laplace matrix L, which is as follows:
[0136] L=IF 1 / 2 ZF -1 / 2
[0137] Among them, I represents the identity matrix, Z represents the comprehensive correlation matrix;
[0138] Calculate the eigenvalues and eigenvectors of the Laplace matrix L, and select the eigenvectors corresponding to the first k smallest eigenvalues to form the eigenvector matrix U, and then perform clustering. Take each row of the eigenvector matrix U as a data point, and use the k-means algorithm to cluster these data points in the feature space. The formula of the k-means is as follows:
[0139]
[0140] Among them, U i represents the i-th row element of the eigenvector matrix U, u j represents the centroid of the jth cluster, k represents the number of clusters, and n represents the number of rows in the eigenvector matrix. The number of clusters k can be determined by the elbow method. By analyzing the numerical changes in the k-means algorithm formula under different cluster numbers, the inflection point position where the numerical value in the k-means algorithm formula drops sharply with the increase of k is determined, and the number of clusters corresponding to the inflection point position is selected as the optimal number of clusters.
[0141] S6. According to the regional division results, select the node with the largest sum of comprehensive correlations with all nodes in any area as the dominant node in the area, and install management equipment on the dominant node to complete the power quality management zoning of the distribution network.
[0142] In this embodiment, the dominant node in the region is the node with the highest correlation with all nodes in the region, so the node with the largest sum of correlations with other nodes in the region is the dominant node.
[0143] In this embodiment, Figure 3 and Figure 4 To verify the effectiveness of the method of the present invention, Figure 2 The IEEE33 node distribution network shown in the figure is simulated and verified. 1 =0.2, b 2 = 0.3 and b 3 =0.5 and compared with the prior art, the zoning results are shown in Table 1 and Table 2 respectively. Table 1 is the regional division result of power quality management by the method of the present invention, and Table 2 is the regional division result of power quality management by the prior art.
[0144] Table 1
[0145]
[0146]
[0147] Table 2
[0148]
[0149] In this embodiment, according to Table 2 and Figure 4It can be seen that the power quality area division based solely on the change trend is easily affected by chance. Even if two nodes are not actually highly correlated, they will be divided into the same area because of their similar change trends. Figure 3 It can be seen that by adopting the method of the present invention, the power quality management area is divided from multiple dimensions such as electrical distance, amplitude and change trend, which can effectively avoid the above situation and make the divided areas more reasonable.
[0150] In this embodiment, Figure 5 , Figure 6 as well as Figure 7 As shown, the three figures show the changes in harmonics, three-phase imbalance and voltage deviation of each node after installing a harmonic control device at the dominant node 10 in the harmonic division area 3, installing a three-phase unbalance control device at the dominant node 11 in the three-phase unbalance division area 3, and installing a voltage deviation control device at the dominant node 12 in the voltage deviation division area 3. It can also be seen that installing control equipment at the dominant node in the region has the best effect on improving the power quality of nodes in the same region, while the improvement effect outside the region is weaker, which is consistent with the zoning result, verifying that the method of the present invention can effectively divide different nodes into regions, achieving strong coupling within the region and weak coupling outside the region.
[0151] Example 2
[0152] The embodiment of the present invention provides a distribution network power quality management partition device, such as Figure 8 As shown, the distribution network power quality management zoning device is characterized by comprising:
[0153] Obtaining electrical distance module: used to calculate the reactive voltage sensitivity and active voltage sensitivity of each node in the distribution network using the power flow equation according to the network topology and parameters of the distribution network, and to obtain the electrical distance between each node according to the reactive voltage sensitivity and active voltage sensitivity of each node;
[0154] Dataset building module: used to build a time series measurement dataset using the historical measurement data of harmonics, three-phase unbalance and voltage deviation of each node;
[0155] Correlation analysis module: used to normalize the time series measurement data set using the normalization method, and use the grey correlation to analyze the normalized time series measurement data set, and obtain the change trend correlation and amplitude correlation of each node respectively;
[0156] Comprehensive correlation module: used to determine the comprehensive correlation between nodes according to the correlation of the change trend, amplitude and electrical distance of each node, and integrate the comprehensive correlation between nodes to obtain the comprehensive correlation matrix;
[0157] Spectral clustering module: used to cluster the comprehensive correlation matrix using the spectral clustering algorithm and determine the final number of regional divisions in combination with the elbow method. According to the final number of regional divisions, regional division for harmonic, three-phase unbalance and voltage deviation management is achieved;
[0158] Selection module: According to the regional division results, the node with the largest sum of comprehensive correlations with all nodes in any area is selected as the dominant node in the area, and management equipment is installed on the dominant node to complete the power quality management zoning of the distribution network.
[0159] The electrical distance acquisition module comprises:
[0160] Newton-power flow equation submodule: used to calculate the reactive voltage sensitivity and active voltage sensitivity of each node based on the Newton-power flow equation according to the network topology and parameters of the distribution network;
[0161] Voltage sensitivity submodule: used to obtain the voltage sensitivity between each node based on the reactive voltage sensitivity and active voltage sensitivity of each node using the sensitivity equation;
[0162] Electrical distance submodule: used to calculate the electrical distance between each node based on the voltage sensitivity between each node using the electrical distance formula.
[0163] The association analysis module includes:
[0164] Normalization submodule: used to normalize the historical measurement data of harmonics, three-phase unbalance and voltage deviation of each node according to the time series measurement data set, and map the historical measurement data to the [0,1] interval;
[0165] Change trend correlation submodule: It is used to analyze the change trend correlation of harmonics, three-phase unbalance and voltage deviation of different nodes using grey correlation analysis method according to the normalized time series measurement data set;
[0166] Amplitude correlation submodule: It is used to analyze the amplitude correlation of harmonics, three-phase unbalance and voltage deviation at different nodes using grey correlation analysis method according to the normalized time series measurement data set.
[0167] The spectral clustering module includes:
[0168] Degree matrix submodule: used to calculate the degree matrix according to the comprehensive correlation matrix of harmonics, three-phase unbalance and voltage deviation of each node, and symmetric normalize the degree matrix to obtain the Laplace matrix;
[0169] Eigenvector submodule: used to select the eigenvectors corresponding to the first k smallest eigenvalues according to the eigenvalues and eigenvectors of the Laplace matrix to form an eigenvector matrix;
[0170] K-means submodule: It is used to treat each row of the feature vector matrix as a data point and cluster the data points using the K-means algorithm in the feature space;
[0171] Elbow method submodule: used to determine the number of clusters based on the clustering results using the elbow method, analyze the value changes in the k-means algorithm under different cluster numbers, and determine the inflection point where the value in the k-means algorithm drops sharply as k increases;
[0172] Optimal number of clusters submodule: It is used to select the number of clusters corresponding to the inflection point position as the optimal number of clusters according to the inflection point position, and use the optimal number of clusters as the final number of region divisions.
[0173] In this embodiment, the present application can divide the modules according to the power quality management zoning method of the distribution network. For example, each function can be divided into each module, or two or more functions can be integrated into one module. The above modules can be implemented in the form of hardware or software. It should be noted that the division of modules in the present invention is schematic and is only a logical division. There may be other division methods in actual implementation.
[0174] In this embodiment, in order to realize the principles and beneficial effects of the power quality management zoning method for distribution network, the power quality management zoning device for distribution network includes hardware structures and / or software parts corresponding to the execution of each function. Those skilled in the art should easily realize that, in combination with the schematic modules and method steps described in the embodiments disclosed in the present invention, the present invention can be implemented in the form of hardware and / or a combination of hardware and computer software. Whether a function is executed in a hardware or computer software driven manner depends on the specific application and design constraints of the technical solution. Different methods can be used for each specific application to implement the described function, but such implementation should not be considered to exceed the scope of this application.
[0175] In this embodiment, a computer device is characterized by comprising: one or more processors;
[0176] The processor is used to store one or more programs; when the one or more programs are executed by the one or more processors, the distribution network power quality management zoning method as described above is implemented.
[0177] In this embodiment, a computer-readable storage medium is provided, characterized in that a computer program is stored thereon, and when the computer program is executed, the distribution network power quality management zoning method as described above is implemented.
Claims
1. A method for zoning power quality management in a distribution network, characterized in that: The following steps are involved: S1. According to the network topology and parameters of the distribution network, the reactive voltage sensitivity and active voltage sensitivity of each node in the distribution network are calculated using the power flow equation, and the electrical distance between each node is obtained according to the reactive voltage sensitivity and active voltage sensitivity of each node; S2, construct a time series measurement data set using the historical measurement data of harmonics, three-phase unbalance and voltage deviation at each node; S3, using a normalization method to normalize the time series measurement data set, and using grey correlation to analyze the normalized time series measurement data set, to obtain the change trend correlation and amplitude correlation of each node respectively; S4, determining the comprehensive correlation between each node according to the correlation of the change trend, the amplitude correlation and the electrical distance of each node, and integrating the comprehensive correlation between each node to obtain a comprehensive correlation matrix; S5. Use the spectral clustering algorithm to cluster the comprehensive correlation matrix and combine it with the elbow method to determine the final number of regional divisions. According to the final number of regional divisions, regional division for harmonic, three-phase unbalance and voltage deviation management is achieved; S6. According to the regional division results, select the node with the largest sum of comprehensive correlations with all nodes in any area as the dominant node in the area, and install management equipment on the dominant node to complete the power quality management zoning of the distribution network.
2. The power quality management zoning method for distribution network according to claim 1 is characterized in that: The S1 step is as follows: S101. According to the network topology and parameters of the distribution network and based on the Newton-power flow equation, the reactive voltage sensitivity and the active voltage sensitivity of each node are calculated; S102, obtaining the voltage sensitivity between the nodes using a sensitivity equation according to the reactive voltage sensitivity and the active voltage sensitivity of each node; S103 . Calculate the electrical distance between the nodes using an electrical distance formula according to the voltage sensitivity between the nodes.
3. The method for zoning power quality management of distribution network according to claim 2 is characterized in that: The equation for the voltage sensitivity is as follows: Among them, A ij represents the voltage sensitivity between node i and node j, Represents the voltage amplitude of node j V j Active voltage sensitivity at node i, Represents the voltage amplitude of node i V i Active voltage sensitivity at node j, Represents the voltage amplitude of node j V j Reactive voltage sensitivity to node i, Represents the voltage amplitude of node i V i The reactive voltage sensitivity of node j is: Represents the voltage amplitude of node i V i The partial derivative of ΔP i represents the change in active power at node i, ΔQ i Represents the change in reactive power at node i.
4. The method for zoning power quality management of distribution network according to claim 1 is characterized in that: The S3 steps are as follows: S301, normalizing the historical measurement data of harmonics, three-phase unbalance and voltage deviation of each node according to the time series measurement data set, and mapping the historical measurement data to the [0,1] interval; S302, using the grey correlation analysis method to analyze the correlation of the changing trends of harmonics, three-phase imbalance and voltage deviation at different nodes according to the normalized time series measurement data set; S303 , according to the normalized time series measurement data set, the grey correlation analysis method is used to analyze the amplitude correlation of harmonics, three-phase unbalance and voltage deviation of different nodes.
5. The method for zoning power quality management of distribution network according to claim 1 is characterized in that: In the step S5, the spectral clustering algorithm is used to cluster the comprehensive correlation matrix and the elbow method is combined to determine the final number of regional divisions, including the following: S501, using a spectral clustering algorithm to divide the distribution network into regions, calculating a degree matrix based on a comprehensive correlation matrix of harmonics, three-phase imbalance, and voltage deviation at each node, and symmetric normalizing the degree matrix to obtain a Laplace matrix; S502, according to the eigenvalues and eigenvectors of the Laplace matrix, selecting the eigenvectors corresponding to the first k smallest eigenvalues to form an eigenvector matrix; S503, taking each row of the eigenvector matrix as a data point, and clustering the data points using the k-means algorithm in the eigenspace; S504, according to the clustering results, using the elbow method to determine the number of clusters, analyzing the value changes in the k-means algorithm under different cluster numbers, and determining the inflection point position where the value in the k-means algorithm drops sharply as k increases; S505 . According to the inflection point position, the number of clusters corresponding to the inflection point position is selected as the optimal number of clusters, and the optimal number of clusters is used as the final number of region divisions.
6. The method for zoning power quality management of distribution network according to claim 5 is characterized in that: The degree matrix F is a diagonal matrix, and the diagonal elements F ii is the sum of the comprehensive associations between all nodes connected to node i, and the diagonal element F ii The formula is as follows: Among them, Z ij Represents the element in the i-th row and j-th column of the comprehensive correlation matrix; The formula of the Laplace matrix L is as follows: L=I-F 1 / 2 ZF -1 / 2 Among them, I represents the identity matrix and Z represents the comprehensive correlation matrix.
7. The method for zoning power quality management of distribution network according to claim 1 is characterized in that: The calculation formula of the comprehensive correlation is as follows: Among them, ε represents the comprehensive correlation, Indicates the correlation degree of harmonic amplitude of each node, Indicates the correlation degree of harmonic change trend of each node, D h represents the electrical distance, b1, b2, and b3 all represent the weight coefficients of amplitude correlation, change trend correlation, and electrical distance, and b1+b2+b3=1; The comprehensive correlation matrix is as follows: Among them, ε ij Represents the comprehensive correlation between the i-th node and the j-th node.
8. A distribution network power quality management zoning device, characterized in that: include: Obtaining electrical distance module: According to the network topology and parameters of the distribution network, the reactive voltage sensitivity and active voltage sensitivity of each node in the distribution network are calculated using the power flow equation, and the electrical distance between each node is obtained based on the reactive voltage sensitivity and active voltage sensitivity of each node; Dataset building module: used to build a time series measurement dataset using the historical measurement data of harmonics, three-phase unbalance and voltage deviation of each node; Correlation analysis module: normalize the time series measurement data set using the normalization method, and analyze the normalized time series measurement data set using the grey correlation degree to obtain the change trend correlation degree and amplitude correlation degree of each node respectively; Comprehensive correlation module: used to determine the comprehensive correlation between nodes according to the correlation of the change trend, amplitude correlation and electrical distance of each node, and integrate the comprehensive correlation between nodes to obtain the comprehensive correlation matrix; Spectral clustering module: used to cluster the comprehensive correlation matrix using the spectral clustering algorithm and determine the final number of regional divisions in combination with the elbow method. According to the final number of regional divisions, regional division for harmonic, three-phase unbalance and voltage deviation management is achieved; Selection module: According to the regional division results, the node with the largest sum of comprehensive correlations with all nodes in any area is selected as the dominant node in the area, and management equipment is installed on the dominant node to complete the power quality management zoning of the distribution network.
9. A distribution network power quality management zoning device according to claim 8, characterized in that: The electrical distance acquisition module comprises: Newton-power flow equation submodule: used to calculate the reactive voltage sensitivity and active voltage sensitivity of each node based on the Newton-power flow equation according to the network topology and parameters of the distribution network; Voltage sensitivity submodule: used to obtain the voltage sensitivity between each node based on the reactive voltage sensitivity and active voltage sensitivity of each node using the sensitivity equation; Electrical distance submodule: used to calculate the electrical distance between each node based on the voltage sensitivity between each node using the electrical distance formula.
10. A distribution network power quality management zoning device according to claim 8, characterized in that: The association analysis module includes: Normalization submodule: used to normalize the historical measurement data of harmonics, three-phase unbalance and voltage deviation of each node according to the time series measurement data set, and map the historical measurement data to the [0,1] interval; Change trend correlation submodule: It is used to analyze the change trend correlation of harmonics, three-phase unbalance and voltage deviation of different nodes using grey correlation analysis method according to the normalized time series measurement data set; Amplitude correlation submodule: It is used to analyze the amplitude correlation of harmonics, three-phase unbalance and voltage deviation at different nodes using grey correlation analysis method according to the normalized time series measurement data set.
11. A distribution network power quality management zoning device according to claim 8, characterized in that: The spectral clustering module includes: Degree matrix submodule: used to calculate the degree matrix according to the comprehensive correlation matrix of harmonics, three-phase unbalance and voltage deviation of each node, and symmetric normalize the degree matrix to obtain the Laplace matrix; Eigenvector submodule: used to select the eigenvectors corresponding to the first k smallest eigenvalues according to the eigenvalues and eigenvectors of the Laplace matrix to form an eigenvector matrix; K-means submodule: It is used to treat each row of the feature vector matrix as a data point and cluster the data points using the K-means algorithm in the feature space; Elbow method submodule: used to determine the number of clusters based on the clustering results using the elbow method, analyze the value changes in the k-means algorithm under different cluster numbers, and determine the inflection point where the value in the k-means algorithm drops sharply as k increases; Optimal number of clusters submodule: It is used to select the number of clusters corresponding to the inflection point position as the optimal number of clusters according to the inflection point position, and use the optimal number of clusters as the final number of region divisions.
12. A computer device, characterized in that: include: one or more processors; The processor is used to store one or more programs; When the one or more programs are executed by the one or more processors, the distribution network power quality management zoning method as described in any one of claims 1 to 7 is implemented.
13. A computer-readable storage medium, characterized in that: A computer program is stored thereon, and when the computer program is executed, the distribution network power quality management zoning method as described in any one of claims 1 to 7 is implemented.
Citation Information
Patent Citations
Optimal selection method for power quality energy-saving and efficiency-improving scheme of power distribution network
CN110707685A
Power distribution network voltage partition control method based on sensitivity clustering analysis
CN116760046A
Self-discipline coordination control method for reactive voltage of power distribution network
CN116937606A
Distribution network harmonic wave and voltage comprehensive governance partition dominant governance node selection method
CN118399407A
Cited By
Power quality zoning treatment method based on time sequence and data driving
CN120834572A
A power quality partitioned treatment method based on time series and data driving
CN120834572B
