A line error parameter identification and estimation method considering power flow distribution of power transmission looped network

Abstract

The application discloses a line error parameter identification and estimation method considering power flow distribution of a power transmission looped network, comprising the following steps: performing network hierarchical topological search based on a minimum principle of node power imbalance, and converting the looped network into a radial structure; analyzing typical features of line parameter errors in the looped network, including concealment under a node power balance criterion, concealment under a consistency criterion of first and last ends of a branch, and abnormal aggregation features of looped network associated branch power residuals; based on the typical features, performing abnormal area preliminary screening through branch power residual tracking, screening a suspected branch by using residual sensitivity analysis, verifying the deviation by combining first and last end power flow consistency, and locating a parameter error line; based on residual sensitivity relationship, partitioning the whole network, simplifying an external network, performing partitioned augmented state estimation on the simplified network, and completing error parameter correction. The application can realize accurate identification and efficient estimation of line error parameters in a power transmission looped network.

Description

Technical Field

[0001] This invention relates to the field of power grid state estimation technology, specifically a method for identifying and estimating line error parameters considering the power flow distribution of transmission ring networks. Background Technology

[0002] In the actual operation of power transmission networks, the accuracy of line impedance parameters is fundamental to the reliable operation of state estimation. When line parameters are incorrect, a deviation will occur between the state estimation model and the actual power grid, leading to a persistently large estimation residual, ultimately resulting in a decrease in the state estimation pass rate. In radial networks, the impact of parameter errors is usually limited to the erroneous line itself and its adjacent area, and the effect is relatively straightforward. However, in transmission networks with ring network structures, when multiple lines in the ring network form parallel paths, the power distribution of each line is determined by the impedance ratio of all parallel lines. When the impedance of one line is incorrect, the power distribution relationship is disturbed as a whole. Not only does the power flow of the erroneous line itself deviate from the true value, but the power of other lines in parallel with it will also change accordingly. This error propagation effect is particularly significant under heavy load conditions because when the line transmission power is large, the absolute value of the power deviation caused by the impedance deviation is also amplified, causing the residuals of multiple related lines in the ring network to exceed the threshold simultaneously.

[0003] Regarding the identification and estimation of parameter errors, existing research has two main representative methods. The first is the maximum standardized residual method. Its basic idea is to calculate the standardized residuals of each measurement after the state estimation converges, and then identify the measurement with the largest absolute residual value as an anomaly and remove or correct it. This method is computationally simple, mature, and performs well in identifying individual defective measurement data. However, its core criterion is based on a threshold comparison of single-point residuals. When the parameters of a line in a ring network are incorrect, this parameter error simultaneously affects the residuals of multiple related lines through the power allocation mechanism. This means that the measurement with the largest residual may not be the measurement of the incorrect line itself, but rather the measurement of a neighboring branch most significantly affected by the propagation effect. This leads to the method easily misjudging residual anomalies caused by parameter errors as bad measurement data, resulting in a bias in the identification direction. The second method is residual sensitivity analysis. Its basic idea is to establish a sensitivity relationship between measurement residuals and line parameters, and to screen suspicious parameter branches by analyzing the degree of influence of changes in the parameters of each branch on the residuals. Compared to the first type of method, this method can more specifically shift the analysis focus from measurement data to network parameters, giving it an advantage in narrowing down the range of suspects. However, when the sensitivity index values ​​of multiple branches in a ring network are similar, it is difficult to make a definitive judgment based on a single sensitivity index. In addition, after completing the identification, both of the above methods usually use full-network augmented state estimation to correct erroneous parameters, but the computational scale of full-network augmentation is large, and the dimension of the Jacobian matrix increases significantly, adding unnecessary computational burden. Summary of the Invention

[0004] The purpose of this invention is to provide a method for identifying and estimating line error parameters considering the power flow distribution of a transmission ring network, comprising the following steps:

[0005] Acquire the network topology of the power transmission network and power grid measurement data;

[0006] Based on the principle of minimizing node power imbalance, a hierarchical topology search is performed on the network topology to determine the loop unblocking point, and the ring network in the network topology is transformed into a radial structure to obtain the branch hierarchy matrix and the branch head and end node matrix.

[0007] Based on the power grid measurement data and state estimation results, the power residual difference index of each branch is calculated;

[0008] Based on the power residual difference index, abnormal areas of the transmission network are initially screened to identify first-level suspicious areas;

[0009] Within the first-level suspected area, a sensitivity matrix based on the measurement residual to the line impedance parameter is constructed, and the relative influence factor of each branch is calculated. Second-level suspected branches are then selected based on the relative influence factor.

[0010] Calculate the deviation in the power flow consistency estimation of the first and last ends of the suspected secondary branches, and locate the erroneous lines by combining the relative influence factor parameters.

[0011] The transmission network is partitioned based on the residual sensitivity relationship. Local networks that are strongly correlated with lines with parameter errors are classified as internal regions, and remote networks that are weakly correlated with lines with parameter errors are classified as external regions.

[0012] The external region is simplified using the extended Ward equivalence method. Augmented state estimation is then performed on the simplified network after equivalence to obtain the correction values ​​for the error parameters.

[0013] Furthermore, the steps for performing a hierarchical topology search on the network topology based on the principle of minimizing node power imbalance include:

[0014] Step 1. Form the node-branch association matrix A based on the node and branch connection information;

[0015] Step 2. Find all nodes with a degree of 1 in the node-branch association matrix A, mark the branches connected to these nodes with a degree of 1 as branches of the same level in the radial network, write them into the branch level matrix L, and remove the nodes and branches from the association matrix A.

[0016] Step 3. Repeat step 2 until there are no more nodes with a degree of 1;

[0017] If there are still unmarked nodes at this point, proceed to step 4;

[0018] Step 4. Calculate the power imbalance of all remaining nodes, and determine the unloop point according to the principle of minimizing the node power imbalance;

[0019] All branches connected to the unloop point are marked and removed as branches at the same level.

[0020] Step 5. Repeat step 4 until all nodes are marked and the association matrix A is a zero matrix;

[0021] The power imbalance at node j is shown below:

[0022]

[0023] in, , This represents the imbalance between active and reactive power at node j. , This represents the measured values ​​of active and reactive power injected into node j; This represents the nodes that are directly connected to node j, excluding node j itself. , This represents the measured active and reactive power values ​​of the branch jk directly connected to node j. This represents the complex power imbalance at node j.

[0024] Furthermore, the step of initially screening abnormal areas of the transmission network based on the power residual difference index is as follows:

[0025] Calculate the power residual difference index of each branch. If the power residual difference index of multiple consecutive electrically adjacent branches exceeds the preset threshold of 0.10, the connected part containing these consecutive abnormal branches will be marked as a first-level suspicious area.

[0026] The power residual difference indexes for each branch are shown below:

[0027]

[0028] in, , This represents the estimated active and reactive power at the beginning of the branch; , This represents the estimated active and reactive power at the end of the branch. The power residual difference index represents the power residual difference of branch ij; , This represents the measured values ​​of active and reactive power in branch ij; , This indicates the measured values ​​of active and reactive power of branch ji.

[0029] Furthermore, the steps for calculating the relative influence factor of each branch include:

[0030] Based on the linear approximation relationship between the measurement residual and the line impedance parameters, a residual sensitivity matrix is ​​constructed. ,Right now:

[0031]

[0032] Where H represents the standard measurement Jacobian matrix; W represents the measurement weight matrix; This represents the matrix of partial derivatives of the measurement function with respect to the parameters; Represents a unit vector;

[0033] Based on residual sensitivity matrix Calculate the relative impact factor ,Right now:

[0034]

[0035] in Indicates relative impact factor; Indicates the area The set of all measurements with non-compliant residuals; This represents the sum of the changes in the residual power measurement of the i-th branch when the parameters of line l change by a unit. Represents the weighted term of the residual; This represents the maximum value of the influence factor of all possible branches;

[0036] The relative impact factors are sorted by size, and the branches in the top N (N=10) are designated as secondary suspicious branches.

[0037] Furthermore, the parameter error branch refers to the branch with the largest joint judgment index;

[0038] The joint judgment indicators are as follows:

[0039]

[0040]

[0041] In the formula, This represents the relative estimation deviation; , These represent the power values ​​for different branches; , Denotes the weight coefficients, and satisfies . These are joint judgment indicators.

[0042] Furthermore, the steps for partitioning the entire network based on residual sensitivity relationships include:

[0043] If the overall sensitivity of a branch is greater than or equal to the sensitivity threshold, then the first and last nodes of the branch are classified into the inner region; if the overall sensitivity of a branch is less than the sensitivity threshold and is directly connected to a node in the inner region, then the node is defined as a boundary node; nodes with an overall sensitivity less than the sensitivity threshold are classified into the outer region.

[0044] Adjustments were made to the preliminary zoning results, including: ensuring the integrity of the ring network selected in the initial screening of abnormal areas, adding buffer layer nodes outside the boundaries, and verifying the observability of the internal areas.

[0045] Furthermore, the step of simplifying the external region using the extended Ward equivalence method includes:

[0046] The admittance matrix of all network nodes is divided into blocks according to internal regions, boundaries, and external regions;

[0047] External nodes are eliminated by Gaussian elimination to obtain the equivalent admittance matrix and equivalent injected current at the boundary nodes;

[0048] Eliminate the external PQ nodes and retain the external PV nodes, and calculate the equivalent admittance between the boundary nodes and the external PV nodes;

[0049] Virtual PV nodes are set at the boundary nodes to maintain the voltage support characteristics of the external generator; the virtual PV nodes do not provide active power in the ground state, and the voltage amplitude of the virtual PV nodes is the same as that of the corresponding boundary nodes.

[0050] Furthermore, the steps for performing augmented state estimation on the simplified network after equivalence include:

[0051] The impedance parameters of the faulty line are extended to state variables. ; These are the regular state variables, including the voltage phase angle vector θ and voltage magnitude vector U of each node; p is the impedance parameter vector of the line with parameter errors.

[0052] Constructing an augmented state estimation model on the simplified network The estimated values ​​of the error parameters are obtained through iterative solution; where z is the measurement vector of the simplified network; To augment state variables is the nonlinear measurement function of the independent variable; W is the measurement weight matrix;

[0053] Replace the erroneous parameter values ​​in the database with the estimated values, and re-evaluate the normal state of the entire network to restore the pass rate.

[0054] Furthermore, assume that the internal region, including boundary nodes, has a total of There are n nodes, and the number of parameters to be estimated is 1. Number of measurements in the internal area satisfy ,in, The dimension representing the regular state variables; Indicates the number of augmented parameter variables.

[0055] A line error parameter identification and estimation device based on the method, considering the power flow distribution of a transmission ring network, includes:

[0056] The topology search module is used to obtain the network topology of the power transmission network and the power grid measurement data. Based on the principle of minimizing the node power imbalance, it performs a hierarchical topology search on the network topology, transforms the ring network into a radial structure, and obtains the branch hierarchy matrix and the branch head and end node matrix.

[0057] The abnormal area screening module is used to calculate the power residual difference index of each branch based on the power grid measurement data and state estimation results, to screen the abnormal areas and identify first-level suspicious areas.

[0058] The suspected branch screening module is used to construct a residual sensitivity matrix in the first-level suspected area, calculate the relative influence factor of each branch, and screen out the second-level suspected branches.

[0059] The error line location module is used to calculate the head-end power flow consistency estimation deviation of the secondary suspected branch, construct a joint judgment index by combining the relative influence factor, and locate the error line parameter.

[0060] The partition estimation module is used to partition the entire network based on the residual sensitivity relationship, simplify the external region using the extended Ward equivalent method, perform augmented state estimation on the simplified network, and obtain the correction values ​​of the error parameters.

[0061] This invention, through in-depth analysis of three typical characteristics of line parameter errors in ring networks—namely, the concealment under the node power balance criterion, the concealment under the branch head-end consistency criterion, and the abnormal clustering of power residuals among multiple associated branches—designs a feature-driven multi-level joint identification strategy. This strategy uses residual tracking for initial screening of abnormal areas, residual sensitivity analysis to filter suspected branches, and then verifies the results by combining head-end power flow consistency calculation deviations, overcoming the limitation of insufficient identification capability of a single criterion in ring network scenarios. In the parameter estimation stage, it uses the extended Ward equivalent method for partitioned augmented state estimation, effectively reducing the computational scale while maintaining estimation accuracy, providing an efficient and practical method for parameter correction in large-scale complex networks. Attached Figure Description

[0062] Figure 1 This is a flowchart of a method for identifying and estimating line error parameters considering the power flow distribution of a transmission ring network, provided by an embodiment of the present invention.

[0063] Figure 2 This is a schematic diagram of a network hierarchical topology search method based on the principle of minimizing node power imbalance provided in an embodiment of the present invention;

[0064] Figure 3 This is a schematic diagram of the concealment analysis under the node power balance criterion in a simple ring network provided by an embodiment of the present invention;

[0065] Figure 4 This is a schematic diagram of a parallel ring network with n branches provided according to an embodiment of the present invention;

[0066] Figure 5 This is a schematic diagram of the extended Ward equivalent model provided according to an embodiment of the present invention;

[0067] Figure 6 This is a schematic diagram of the residual distribution under different reactance errors according to an embodiment of the present invention;

[0068] Figure 7 This is a schematic diagram comparing the recognition accuracy of different recognition methods provided in the embodiments of the present invention;

[0069] Figure 8 This is a schematic diagram comparing the pass rate of the entire network state estimation before and after parameter correction according to an embodiment of the present invention;

[0070] Figure 9 This is a structural block diagram of a line error parameter identification and estimation device that considers the power flow distribution of a transmission ring network, according to an embodiment of the present invention. Detailed Implementation

[0071] The present invention will be further described below with reference to embodiments, but it should not be construed that the scope of the present invention is limited to the following embodiments. Various substitutions and modifications made based on ordinary technical knowledge and common practices in the art without departing from the above-described technical concept of the present invention should be included within the scope of protection of the present invention.

[0072] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion.

[0073] Before introducing the embodiments of the present invention, the influence mechanism of line parameter errors in the power transmission ring network will be explained by way of example in order to better understand the solution proposed in the embodiments of the present invention.

[0074] For example, in a transmission network with a ring network structure, when multiple lines in the ring network form parallel paths, the power distribution of each line is determined by the impedance ratio of all parallel lines. When the impedance of a certain line is incorrect, the power distribution relationship is disturbed as a whole. Not only does the power flow of the incorrect line deviate from the true value, but the power of the lines in other parallel paths will also change accordingly. Since parameter errors change the model itself rather than the measurement data, traditional bad data identification methods based on the consistency of measurement data cannot effectively identify this type of error. This makes the existence of line parameter errors in the ring network a hidden factor restricting the improvement of the pass rate.

[0075] Example 1:

[0076] See Figures 1-9 A method for identifying and estimating line error parameters considering the power flow distribution of a transmission ring network includes the following steps:

[0077] Acquire the network topology of the power transmission network and power grid measurement data;

[0078] Based on the principle of minimizing node power imbalance, a hierarchical topology search is performed on the network topology to determine the loop unblocking point, and the ring network in the network topology is transformed into a radial structure to obtain the branch hierarchy matrix and the branch head and end node matrix.

[0079] Based on the power grid measurement data and state estimation results, the power residual difference index of each branch is calculated;

[0080] Based on the power residual difference index, abnormal areas of the transmission network are initially screened to identify first-level suspicious areas;

[0081] Within the first-level suspected area, a sensitivity matrix based on the measurement residual to the line impedance parameter is constructed, and the relative influence factor of each branch is calculated. Second-level suspected branches are then selected based on the relative influence factor.

[0082] Calculate the deviation in the power flow consistency estimation of the first and last ends of the suspected secondary branches, and locate the erroneous lines by combining the relative influence factor parameters.

[0083] The transmission network is partitioned based on the residual sensitivity relationship. Local networks that are strongly correlated with lines with parameter errors are classified as internal regions, and remote networks that are weakly correlated with lines with parameter errors are classified as external regions.

[0084] The external region is simplified using the extended Ward equivalence method. Augmented state estimation is then performed on the simplified network after equivalence to obtain the correction values ​​for the error parameters.

[0085] Example 2:

[0086] A method for identifying and estimating line error parameters considering the power flow distribution of a transmission ring network, with the same technical content as in Embodiment 1, further comprising the step of performing a hierarchical topology search on the network topology based on the principle of minimizing node power imbalance, including:

[0087] Step 1. Form the node-branch association matrix A based on the node and branch connection information;

[0088] Step 2. Find all nodes with a degree of 1 in the node-branch association matrix A, mark the branches connected to these nodes with a degree of 1 as branches of the same level in the radial network, write them into the branch level matrix L, and remove the nodes and branches from the association matrix A.

[0089] Step 3. Repeat step 2 until there are no more nodes with a degree of 1;

[0090] If there are still unmarked nodes at this point, proceed to step 4;

[0091] Step 4. Calculate the power imbalance of all remaining nodes, and determine the unloop point according to the principle of minimizing the node power imbalance;

[0092] All branches connected to the unloop point are marked and removed as branches at the same level.

[0093] Step 5. Repeat step 4 until all nodes are marked and the association matrix A is a zero matrix;

[0094] The power imbalance at node j is shown below:

[0095]

[0096] in, , This represents the imbalance between active and reactive power at node j. , This represents the measured values ​​of active and reactive power injected into node j; This represents the nodes that are directly connected to node j, excluding node j itself. , This represents the measured active and reactive power values ​​of the branch jk directly connected to node j. This represents the complex power imbalance at node j.

[0097] Example 3:

[0098] A method for identifying and estimating line error parameters considering the power flow distribution of a transmission ring network, with the same technical content as any one of Embodiments 1-2, further comprising the following steps for preliminary screening of abnormal areas of the transmission network based on the power residual difference index:

[0099] Calculate the power residual difference index of each branch. If the power residual difference index of multiple consecutive electrically adjacent branches exceeds the preset threshold of 0.10, the connected part containing these consecutive abnormal branches will be marked as a first-level suspicious area.

[0100] The power residual difference indexes for each branch are shown below:

[0101]

[0102] in, , This represents the estimated active and reactive power at the beginning of the branch; , This represents the estimated active and reactive power at the end of the branch. The power residual difference index represents the power residual difference of branch ij; , This represents the measured values ​​of active and reactive power in branch ij; , This indicates the measured values ​​of active and reactive power of branch ji.

[0103] Example 4:

[0104] A method for identifying and estimating line error parameters considering the power flow distribution of a transmission ring network, with the same technical content as any one of embodiments 1-3, further comprising the step of calculating the relative influence factor of each branch including:

[0105] Based on the linear approximation relationship between the measurement residual and the line impedance parameters, a residual sensitivity matrix is ​​constructed. ,Right now:

[0106]

[0107] Where H represents the standard measurement Jacobian matrix; W represents the measurement weight matrix; This represents the matrix of partial derivatives of the measurement function with respect to the parameters; Represents a unit vector;

[0108] Based on residual sensitivity matrix Calculate the relative impact factor ,Right now:

[0109]

[0110] in Indicates relative impact factor; Indicates the area The set of all measurements with non-compliant residuals; This represents the sum of the changes in the residual power measurement of the i-th branch when the parameters of line l change by a unit. Represents the weighted term of the residual; This represents the maximum value of the influence factor of all possible branches;

[0111] The relative impact factors are sorted by size, and the branches in the top N (N=10) are designated as secondary suspicious branches.

[0112] Example 5:

[0113] A method for identifying and estimating line error parameters considering the power flow distribution of a transmission ring network, with the same technical content as any one of embodiments 1-4, further wherein the parameter error branch refers to the branch with the largest joint judgment index;

[0114] The joint judgment indicators are as follows:

[0115]

[0116]

[0117] In the formula, This represents the relative estimation deviation; , These represent the power values ​​for different branches; , Denotes the weight coefficients, and satisfies .

[0118] Example 6:

[0119] A method for identifying and estimating line error parameters considering the power flow distribution of a transmission ring network, with technical content identical to any one of embodiments 1-5, further comprising the step of partitioning the entire network based on residual sensitivity relationships, including:

[0120] If the overall sensitivity of a branch is greater than or equal to the sensitivity threshold, then the first and last nodes of the branch are classified into the inner region; if the overall sensitivity of a branch is less than the sensitivity threshold and is directly connected to a node in the inner region, then the node is defined as a boundary node; nodes with an overall sensitivity less than the sensitivity threshold are classified into the outer region.

[0121] Adjustments were made to the preliminary zoning results, including: ensuring the integrity of the ring network selected in the initial screening of abnormal areas, adding buffer layer nodes outside the boundaries, and verifying the observability of the internal areas.

[0122] Example 7:

[0123] A method for identifying and estimating line error parameters considering the power flow distribution of a transmission ring network, with technical content identical to any one of embodiments 1-6, further comprising the step of simplifying the external region using the extended Ward equivalent method, including:

[0124] The admittance matrix of all network nodes is divided into blocks according to internal regions, boundaries, and external regions;

[0125] External nodes are eliminated by Gaussian elimination to obtain the equivalent admittance matrix and equivalent injected current at the boundary nodes;

[0126] Eliminate the external PQ nodes and retain the external PV nodes, and calculate the equivalent admittance between the boundary nodes and the external PV nodes;

[0127] Virtual PV nodes are set at the boundary nodes to maintain the voltage support characteristics of the external generator; the virtual PV nodes do not provide active power in the ground state, and the voltage amplitude of the virtual PV nodes is the same as that of the corresponding boundary nodes.

[0128] Example 8:

[0129] A method for identifying and estimating line error parameters considering the power flow distribution of a transmission ring network, with the same technical content as any one of embodiments 1-7, further comprising the step of performing augmented state estimation on the simplified network after equivalence, including:

[0130] The impedance parameters of the faulty line are extended to state variables. ; These are the regular state variables, including the voltage phase angle vector θ and voltage magnitude vector U of each node; p is the impedance parameter vector of the line with parameter errors.

[0131] Constructing an augmented state estimation model on the simplified network The estimated values ​​of the error parameters are obtained through iterative solution; where z is the measurement vector of the simplified network; To augment state variables is the nonlinear measurement function of the independent variable; W is the measurement weight matrix;

[0132] Replace the erroneous parameter values ​​in the database with the estimated values, and re-evaluate the normal state of the entire network to restore the pass rate.

[0133] Example 9:

[0134] A method for identifying and estimating line error parameters considering the power flow distribution of a transmission ring network, with technical content identical to any one of embodiments 1-8, further assuming that the internal region includes boundary nodes with a total of There are n nodes, and the number of parameters to be estimated is 1. Number of measurements in the internal area satisfy ,in, The dimension representing the regular state variables; Indicates the number of augmented parameter variables.

[0135] Example 10:

[0136] A line error parameter identification and estimation device considering the power flow distribution of a transmission ring network, based on the method described in any one of Embodiments 1-9, includes:

[0137] The topology search module is used to obtain the network topology of the power transmission network and the power grid measurement data. Based on the principle of minimizing the node power imbalance, it performs a hierarchical topology search on the network topology, transforms the ring network into a radial structure, and obtains the branch hierarchy matrix and the branch head and end node matrix.

[0138] The abnormal area screening module is used to calculate the power residual difference index of each branch based on the power grid measurement data and state estimation results, to screen the abnormal areas and identify first-level suspicious areas.

[0139] The suspected branch screening module is used to construct a residual sensitivity matrix in the first-level suspected area, calculate the relative influence factor of each branch, and screen out the second-level suspected branches.

[0140] The error line location module is used to calculate the head-end power flow consistency estimation deviation of the secondary suspected branch, construct a joint judgment index by combining the relative influence factor, and locate the error line parameter.

[0141] The partition estimation module is used to partition the entire network based on the residual sensitivity relationship, simplify the external region using the extended Ward equivalent method, perform augmented state estimation on the simplified network, and obtain the correction values ​​of the error parameters.

[0142] Example 11:

[0143] A method for identifying and estimating line error parameters considering power flow distribution in transmission ring networks, comprising the following steps:

[0144] The network topology of the power transmission network and power grid measurement data are obtained. Based on the principle of minimizing node power imbalance, a hierarchical topology search is performed on the network topology to determine the loop point and transform the ring network into a radial structure, thereby obtaining the branch hierarchy matrix and the branch head and end node matrix.

[0145] Based on the power grid measurement data and state estimation results, the power residual difference index of each branch is calculated, and the abnormal areas are initially screened according to the power residual difference index to determine the first-level suspicious areas.

[0146] Within the first-level suspected area, a sensitivity matrix of the measurement residual to the line impedance parameter is constructed, the relative influence factor of each branch is calculated, and the second-level suspected branches are screened out.

[0147] The deviation in the power flow consistency estimation of the first and last ends of the suspected secondary branches is calculated, and a joint judgment index is constructed in combination with the relative influence factor to locate the line with incorrect parameters.

[0148] The entire network is partitioned based on the residual sensitivity relationship. Local networks strongly correlated with faulty lines are classified as internal regions, while weakly correlated remote networks are classified as external regions. The external regions are simplified using the extended Ward equivalence method. Augmented state estimation is then performed on the simplified network to obtain the correction values ​​for the faulty parameters.

[0149] The hierarchical topology search of the network topology based on the principle of minimizing node power imbalance includes:

[0150] A node-branch association matrix A is formed based on the node and branch connection information;

[0151] Find all nodes with a degree of 1 in the node-branch association matrix A, mark the branches connected to these nodes as branches of the same level in the radial network, write them into the branch level matrix L, and remove the nodes and branches from the association matrix A.

[0152] Repeat the above operation until there are no more nodes with a degree of 1;

[0153] If there are still unmarked nodes, calculate the power imbalance for all remaining nodes, and determine the loop-breaking point according to the principle of minimizing node power imbalance. The formula for calculating the power imbalance of node j is: ,in, , This represents the imbalance between active and reactive power at node j. , This represents the measured values ​​of active and reactive power injected into node j; This represents the nodes that are directly connected to node j, excluding node j itself. , This represents the measured active and reactive power values ​​of the branch jk directly connected to node j. This represents the complex power imbalance at node j.

[0154] All branches connected to the unloop point are marked and removed as branches at the same level. Repeat the above steps until all nodes are marked and the correlation matrix A is a zero matrix.

[0155] Calculate the power residual difference index for each branch. ,in, , This represents the estimated active and reactive power at the beginning of the branch; , This represents the estimated active and reactive power at the end of the branch.

[0156] If the power residual difference index of multiple consecutive electrically adjacent branches approaches a preset threshold, the connected portion containing these consecutive abnormal branches will be marked as a first-level suspicious area.

[0157] The calculation of the relative influence factor for each branch includes:

[0158] Establish a linear approximation relationship between the measurement residuals and the line impedance parameters, and construct the residual sensitivity matrix. , Where H represents the standard measurement Jacobian matrix; W represents the measurement weight matrix; This represents the matrix of partial derivatives of the measurement function with respect to the parameters. The relative influence factor is defined. Its calculation formula is ,in Indicates relative impact factor; Indicates the area The set of all measurements with non-compliant residuals; This represents the sum of the variation amplitudes of the power measurement residuals of the i-th branch when the parameters of line l change by a unit. The larger this value is, the more sensitive the measurement is to changes in the parameters of line l, and the greater the possibility that errors in the parameters of line l will cause abnormalities in the measurement residuals. This represents the residual weighting term. The purpose of introducing the actual residual as a weighting factor is to prioritize branches with larger residual amplitudes. If a branch has a large measurement residual and is also highly sensitive to its parameters, then that branch is most likely to have parameter errors. This represents the maximum value of the influence factor of all possible branches.

[0159] The calculation of the joint judgment index based on the deviation in the power flow consistency estimation at the beginning and end of the branch includes:

[0160] For each suspected branch within the first-level suspected area, the measured value at one end of the branch is used as input, and the power calculation value at the other end is calculated based on the branch power flow equation. The relative calculation deviation is then calculated. ;

[0161] Calculate the joint judgment index The calculation formula is: ,in, , Denotes the weight coefficients, and satisfies .

[0162] The joint judgment index The largest branch was determined to be a branch with incorrect parameters.

[0163] The method of partitioning the entire network based on residual sensitivity relationships includes:

[0164] Set a sensitivity threshold and divide the area into zones based on the comprehensive sensitivity of each branch to the impedance parameters of the line with incorrect parameters: if the comprehensive sensitivity of a branch is greater than or equal to the sensitivity threshold, the first and last nodes of the branch are classified into the inner zone; if the comprehensive sensitivity of a branch is less than the sensitivity threshold and is directly connected to the nodes in the inner zone, the node is defined as a boundary node; the remaining nodes are classified into the outer zone.

[0165] Adjustments were made to the preliminary zoning results, including: ensuring the integrity of the ring network selected in the initial screening of abnormal areas, adding buffer layer nodes outside the boundaries, and verifying the observability of the internal areas.

[0166] The process of simplifying the external region using the extended Ward equivalence method includes:

[0167] The admittance matrix of all network nodes is divided into blocks according to internal regions, boundaries, and external regions;

[0168] External nodes are eliminated by Gaussian elimination to obtain the equivalent admittance matrix and equivalent injected current at the boundary nodes;

[0169] Eliminate the external PQ nodes and retain the external PV nodes, and calculate the equivalent admittance between the boundary nodes and the external PV nodes;

[0170] Virtual PV nodes are set at the boundary nodes. These virtual PV nodes do not provide active power in the ground state, and their voltage amplitude is the same as that of the corresponding boundary nodes. They are used to maintain the voltage support characteristics of the external generator.

[0171] The process of performing augmented state estimation on the simplified network after equivalence includes:

[0172] The impedance parameters of the faulty line are extended into state variables;

[0173] An augmented state estimation model is constructed on the simplified network, and the estimated values ​​of the error parameters are obtained by iterative solution.

[0174] Replace the erroneous parameter values ​​in the database with the estimated values, and then re-evaluate the normal status of the entire network to restore the pass rate.

[0175] The partitioned internal region satisfies the observability condition for augmented state estimation, specifically:

[0176] Assume the internal region, including boundary nodes, has a total of There are n nodes, and the number of parameters to be estimated is 1. The number of measurements in the internal area satisfy ,in, The dimension representing the regular state variables; Indicates the number of augmented parameter variables.

[0177] Example 12:

[0178] See Figure 1 A method for identifying and estimating line error parameters considering the power flow distribution of a transmission ring network, comprising the following steps:

[0179] S110. Obtain the network topology of the power transmission network and the power grid measurement data. Based on the principle of minimizing the node power imbalance, perform a hierarchical topology search on the network topology to determine the loop point and transform the ring network into a radial structure, thereby obtaining the branch hierarchy matrix and the branch head and end node matrix.

[0180] The network topology refers to the connection relationships between nodes and branches in the transmission network. Grid measurement data includes measurement information such as node voltage, node injected active power, node injected reactive power, active power at the beginning of a branch, reactive power at the beginning of a branch, active power at the end of a branch, and reactive power at the end of a branch.

[0181] In this embodiment of the invention, when using the forward-backward substitution method to identify defective data, whether it is a simple ring network or a complex ring network, topological hierarchical analysis of the network is required to form a branch hierarchy matrix L and a matrix M containing the first and last nodes of all branches in the network, for subsequent calculation and analysis. Radial networks, due to their relatively simple structure, only require repeated searching based on the principle of node degree being 1 to form the hierarchy matrix during topological analysis. However, simple and complex ring networks cannot satisfy the principle of node degree being 1, therefore, loop decomposition processing is required. Under complete measurement conditions, using the principle of minimizing node power imbalance, loop decomposition points can be determined in simple and complex ring networks, decomposing the ring network into a radial structure.

[0182] Specifically, the power imbalance at node j is calculated as follows: the active power imbalance at node j equals the measured value of the active power injected into node j minus the sum of the measured values ​​of the active power of all branches directly connected to node j; the reactive power imbalance at node j equals the measured value of the reactive power injected into node j minus the sum of the measured values ​​of the reactive power of all branches directly connected to node j. The complex power imbalance at node j is determined by the active power imbalance and the reactive power imbalance. Physically, the power imbalance reflects the degree of agreement between the measured data at the node and Kirchhoff's current law; a smaller imbalance indicates higher measurement reliability.

[0183] The selection of a loop-breaking point should meet the following conditions: a simple ring network has one and only one loop-breaking point; for a complex ring network containing k independent loops, theoretically k loop-breaking points are needed to transform it into a radial structure, and the number of independent loops can be determined by subtracting the number of nodes from the number of branches in the network; the loop-breaking point must be selected from all nodes contained in a certain ring network according to the principle of minimizing the node power imbalance.

[0184] The detailed steps for topology search based on the principles of minimizing node power imbalance and ensuring node degree is 1 are as follows:

[0185] Step 1: Form the node-branch association matrix A based on the node and branch connection situation;

[0186] Step 2: Find all nodes with a degree of 1 in matrix A, i.e., the column containing the node has only one row with 1 and the rest are all 0. Mark the branches connected to these nodes as branches of the same level in the radial network, write the branch number into the same row of the branch level matrix L, and then remove the nodes and branches from the association matrix A.

[0187] Step 3: Repeat the operation of step 2 for matrix A after removing nodes and branches at the same level until there are no nodes with a degree of 1 in matrix A, which means that all radial network branches have been identified.

[0188] Step 4: If there are still unmarked nodes at this point, it means that there is a ring network in the network and these nodes exist in the ring network. Calculate the power imbalance for all remaining nodes and determine the unblocking point by applying the principle of minimizing the node power imbalance.

[0189] Step 5: Identify all branches connected to the loop unblocking point as branches of the same level in the radial network, and then remove the nodes and branches.

[0190] Step six: Repeat steps one through four until all nodes are marked and all branches are identified. At this point, matrix A is a zero matrix.

[0191] After completing the network topology search, this embodiment of the invention analyzes the typical characteristics of line parameter errors in ring networks, including three aspects:

[0192] Feature 1: Concealment under the node power balance criterion. When impedance errors exist in ring network branches, the node power balance condition still holds because the parameter errors do not disrupt the physical consistency between measurement data, and the measurement data is collected based on the actual operating system. This is a key feature that makes erroneous parameters difficult to detect using traditional methods.

[0193] Feature Two: Concealment under Branch-to-Branch Consistency Criteria. In a ring network topology, system power flow is distributed via multiple loop paths. The power distribution on any branch depends not only on the parameters of that branch itself but also on the coupled constraints of the parameters and operating states of other branches in the network. When a parameter error exists on a single branch in the ring network, the consistency verification index at the single-branch level is still within the acceptable threshold, making it difficult to effectively identify such topologically concealed parameter errors.

[0194] Feature 3: Anomaly Clustering of Power Residuals in Linked Branches of Ring Networks. Although erroneous parameters are often concealed by traditional bad data detection methods, they exhibit significant anomalies in residual analysis during state estimation. This is primarily manifested in the simultaneous exceeding of limits in the power residuals of multiple linked branches within the ring network. The residual offset caused by parameter errors is directly proportional to the parameter error itself. When the impedance of a line in the ring network is incorrect, not only will the power residual of that line itself be abnormal, but the power residuals of multiple branches that are electrically close to and highly coupled with the line with the erroneous parameter will also show anomalies. Furthermore, the direction of power change in the line with the erroneous parameter is opposite to the direction of power change in other branches.

[0195] S120. Based on the power grid measurement data and state estimation results, calculate the power residual difference index of each branch, and perform preliminary screening of abnormal areas according to the power residual difference index to determine the first-level suspicious areas.

[0196] It should be noted that the state estimation results described in this embodiment of the invention are known data. They are the voltage amplitude and phase angle of each node and the estimated active and reactive power of each branch obtained after performing a conventional state estimation on the power grid measurement data. They are given as input to this step along with the power grid measurement data and do not need to be generated separately.

[0197] In this embodiment of the invention, the abnormal clustering pattern of residuals in multiple associated branches revealed by Feature 3 is used to initially screen abnormal areas. First, the measured values ​​are compared with the estimated values, and the branch power residual difference index is defined as the key indicator for the initial screening of abnormal areas. This index integrates the differences between the measured and estimated active power values ​​at the beginning of the branch, the differences between the measured and estimated reactive power values ​​at the beginning of the branch, the differences between the measured and estimated active power values ​​at the end of the branch, and the differences between the measured and estimated reactive power values ​​at the end of the branch.

[0198] Within a network, if multiple consecutive electrically adjacent branches have power residual difference indices approaching a threshold, the connected portion containing these consecutive abnormal branches is marked as a first-level suspicious area.

[0199] S130. Within the first-level suspected area, construct a sensitivity matrix of the measurement residual to the line impedance parameter, calculate the relative influence factor of each branch, and screen out the second-level suspected branches.

[0200] In this embodiment of the invention, the designated first-level suspicious area contains multiple branches. To further pinpoint the line most likely to have erroneous parameters, a residual sensitivity analysis method is employed. Specifically, there is a linear approximation relationship between measurement residuals and erroneous parameters. A residual sensitivity matrix can be constructed, where each element represents the sensitivity of the i-th measurement residual to the impedance parameter of the l-th branch. This matrix can be calculated using the conventional measurement Jacobian matrix, the measurement weight matrix, and the partial derivative matrix of the measurement function with respect to the parameters.

[0201] It should be noted that the erroneous parameters mentioned in the embodiments of the present invention refer to the impedance parameters of the line (including resistance and reactance); the residual sensitivity relationship refers to the linear approximate relationship between the measurement residual and the line impedance parameters, as represented by the sensitivity matrix based on the measurement residual and the line impedance parameters. The subsequent network partitioning based on the residual sensitivity relationship is based on the comprehensive sensitivity of each branch in the sensitivity matrix to the erroneous line impedance parameters.

[0202] To quantify the impact of each possible line parameter on the overall residual anomaly pattern in the region, a relative influence factor is defined. This factor is calculated as follows: for all measurements with non-compliant residuals within the region, the sum of the changes in the power measurement residuals of each branch when the line parameter changes by a unit is calculated. The actual residual is then introduced as a weighting factor, and the result is normalized by dividing by the maximum value of the influence factors of all possible branches. If a branch has a large measurement residual and is also highly sensitive to its branch parameters, then that branch is most likely to have parameter errors.

[0203] By calculating and ranking the relative impact factors of all branches within the region, a branch with a significantly higher relative impact factor than other branches is identified as a secondary suspected branch. If multiple branches have similar relative impact factor values ​​and are all at a high level, these branches are all listed as secondary suspected branches and proceed to the next verification step.

[0204] S140. Calculate the head-end power flow consistency deviation for the second-level suspected branch, and construct a joint judgment index in combination with the relative influence factor to locate the line with incorrect parameters.

[0205] In this embodiment of the invention, when sensitivity analysis cannot provide a unique determination, it is necessary to introduce a head-end estimation deviation verification method. The theoretical basis of this method is feature two: although the head-end estimation deviation of the line with incorrect parameters does not reach the traditional detection threshold, the deviation value is larger than that of other branches in the ring network.

[0206] Specifically, the relative estimation deviation of each secondary suspected branch is calculated, and then a joint judgment index is constructed. The joint judgment index is equal to the sum of the relative influence factor multiplied by the first weighting coefficient and the relative estimation deviation multiplied by the second weighting coefficient, where the sum of the first and second weighting coefficients equals 1. The branch with the highest joint judgment index is judged as a parameter error line.

[0207] The branch power flow equations upon which the head-end power flow consistency calculation is based are specifically as follows: ,in, This represents the reactive power generated by the i-terminal relative to ground susceptance; Indicates the total line power; Indicates the line current; This represents the voltage loss across the impedance; Indicates power loss across impedance; This indicates the voltage at terminal j of the line; Indicates the total line power; This represents the reactive power generated by the ground susceptance at terminal j; Indicates the percentage of voltage error; Indicates the percentage of active power error; This indicates the percentage of reactive power error.

[0208] S150. Based on the residual sensitivity relationship, the entire network is partitioned, and the external region is simplified using the extended Ward equivalent method. Augmented state estimation is performed on the simplified network to obtain the correction values ​​of the error parameters.

[0209] In this embodiment of the invention, after identifying the faulty lines, it is necessary to accurately estimate the faulty parameters of the located lines. While it is theoretically feasible to directly perform augmented state estimation on the entire network, it suffers from problems such as large computational scale, low contribution from remote measurements, and difficulty in synchronizing large-scale system data.

[0210] (1) Network partitioning method based on residual sensitivity

[0211] The goal of network partitioning is to divide the entire network into three regions: the internal region, the boundary region, and the external region. The internal region contains lines with identified parameter errors, along with their strongly correlated nodes and branches; this is the core area requiring detailed modeling and parameter correction. The boundary region is a set of transitional nodes connecting the internal and external networks. The external region is the network portion with weak correlation to the sensitivity of the erroneous parameters and will be replaced by an equivalent model.

[0212] Specifically, a sensitivity threshold is set, and the regions are determined based on the comprehensive sensitivity of each branch to the impedance parameters of the line with incorrect parameters: if the comprehensive sensitivity of a branch is greater than or equal to the sensitivity threshold, the first and last nodes of the branch are classified into the inner region; if the comprehensive sensitivity of a branch is less than the sensitivity threshold, and the branch is directly connected to at least one node in the inner region, the node is defined as a boundary node; the remaining nodes are classified into the outer region.

[0213] The overall sensitivity of each branch is obtained by normalizing its sensitivity value relative to the impedance parameter of the line with parameter error, and its value range is [0, 1]. The sensitivity threshold needs to be determined by a trade-off between the calculation accuracy and the calculation scale of the internal region: if the threshold is too large, the internal region will be too small, resulting in the loss of measurement information useful for parameter estimation; if the threshold is too small, the internal region will be too large, losing the significance of zoning simplification. In this embodiment, the selectable value range of the sensitivity threshold is 0.01 to 0.05, for example, 0.02. This threshold can be flexibly adjusted according to the scale of different power grids and the degree of ring network coupling without affecting the protection scope of the present invention.

[0214] It needs further clarification that the "remaining nodes are assigned to the outer region" criterion in the above partitioning criteria specifically means that all remaining nodes in the entire network, excluding the first and last nodes of branches already assigned to the internal region and the defined boundary nodes, are assigned to the outer region. In other words, for a branch with a comprehensive sensitivity less than the sensitivity threshold, if its first and last nodes are not directly connected to any internal region node, then the node is assigned to the outer region; if its first and last nodes are directly connected to at least one internal region node, then the node is defined as a boundary node. Thus, the set of internal region nodes, the set of boundary nodes, and the set of external region nodes do not overlap and together constitute the entire network node set. Taking the IEEE 118-node example, with a sensitivity threshold of 0.02, the internal region contains 15 nodes, the boundary nodes are 7, and the outer region contains the remaining 96 nodes.

[0215] The preliminary zoning results were adjusted and optimized as follows: First, ring network integrity constraints were implemented to ensure that the ring networks connected by the lines screened in the initial screening of abnormal areas belonged to the internal regions; second, boundary buffer layers were set up by adding at least one layer of electrical distance buffer nodes outside the initially determined boundaries to reduce the impact of equivalence errors on parameter estimation; third, observability verification was performed, ensuring that the internal regions after zoning met the observability conditions of augmented state estimation, i.e., the number of measurements in the internal regions should be greater than or equal to the dimension of the conventional state variables plus the number of augmented parameter variables.

[0216] The specific implementation process for the above three adjustment methods is as follows:

[0217] ① The implementation method of ring network integrity constraint is as follows: For the ring network connected by each branch in the first-level suspicious area screened by the abnormal area screening step, check whether all the nodes contained in the ring network have been included in the internal area; if there are nodes in the ring network that have not yet been included in the internal area, then these nodes are added to the internal area together, so as to ensure that the same ring network will not be split into the internal area and the external area, and avoid the equivalence error caused by incomplete ring unblocking.

[0218] ② The number and standard for setting buffer layer nodes are as follows: The number of buffer layers is at least one electrical distance, and in this embodiment, it is preferably set to one layer. The setting standard is as follows: traverse each initially determined boundary node, include the nodes directly connected to it in the outer region (i.e., with an electrical distance of 1) into the inner region, and merge all newly included nodes with the original boundary nodes into the inner region. Then, redetermine the new set of boundary nodes according to the aforementioned partitioning criteria. The purpose of setting a buffer layer is to limit the error introduced by the equivalent approximation to the buffer layer range and prevent it from propagating to the core area of ​​parameter estimation in the inner region. Taking the IEEE 118-node example, after adding one buffer layer based on the initial partitioning (15 nodes in the inner region and 7 nodes in the boundary region), the number of nodes in the inner region increases to 22, and the number of boundary nodes increases to 18.

[0219] ③ The method for verifying the observability of the internal region is as follows: Assume the internal region (including boundary nodes) has a total of There are n nodes, and the number of parameters to be estimated is 1. The number of measurements in the internal area Should meet ,in The dimension of the conventional state variables (node ​​voltage magnitude and phase angle, taking one node as the reference phase angle). To increase the number of parameter variables. During verification, count the total number of valid measurements within the internal region (including boundary nodes). ,like If the above inequality is satisfied, the internal region is determined to be observable and augmented state estimation can be performed; if not, the buffer layer nodes are expanded outward in accordance with ② to increase the number of measurements in the internal region until the observability condition is met.

[0220] (2) Extended Ward Equivalence Method

[0221] After completing the network partitioning, the external region is simplified by equivalent representation. The extended Ward equivalent method, based on the basic Ward equivalent, introduces virtual branches and voltage constraints to maintain the voltage support characteristics of the external network, thus more accurately reflecting the impact of the external network on the internal system.

[0222] The core idea of ​​basic Ward's equivalent method is to eliminate external nodes using Gaussian elimination, thus equating the influence of the external network to the boundary nodes. First, the network-wide node admittance matrix is ​​divided into three regions: internal, boundary, and external. From the third line of the network equations, the external node voltages can be expressed as boundary node voltages. After substituting these into the second line of equations to eliminate the external node voltages, the equivalent admittance matrix and equivalent injected current at the boundary nodes can be defined. The physical meaning of the equivalent admittance matrix is ​​the equivalent admittance from the boundary node to the external network; the equivalent injected current includes the equivalent injection of external network load and generation.

[0223] However, basic Ward equivalence has limitations: the voltage regulation capability of generators in the external network is lost during the equivalence process. To overcome this deficiency, extended Ward equivalence introduces the following improvements:

[0224] First, eliminate the external PQ nodes, retaining the external PV nodes, to obtain the equivalent admittance matrix between the boundary nodes and the external PV nodes. Then, sum the branch admittances between the boundary nodes and each external PV node to obtain the equivalent susceptance. Ignoring the parallel branches of the external network, eliminate all external nodes to obtain the equivalent admittance matrix between the boundary nodes.

[0225] Virtual PV nodes are set at the boundary nodes. These virtual PV nodes do not provide active power in the ground state, and their voltage amplitude is the same as that of the corresponding boundary nodes. When the system is running, reactive power regulation is automatically performed to maintain the boundary voltage level, thereby preserving the voltage support characteristics of the external PV nodes.

[0226] (3) Augmented state estimation

[0227] On the simplified network after equivalence, the impedance parameters of the faulty lines are augmented as state variables to construct an augmented state estimation model. Estimates of the faulty parameters are obtained through iterative solving. These estimates replace the faulty parameter values ​​in the database, and a new conventional state estimation is performed on the entire network to restore the pass rate.

[0228] The specific implementation process of the above augmented state estimation is as follows:

[0229] The first step is to construct augmented state variables. The augmented state variables... It is constructed by adding the impedance parameter of the faulty line as a variable to be estimated, based on the conventional state variables. ,in For regular state variables, there are voltage phase angle vectors θ and voltage magnitude vectors U for each node in the simplified network (including boundary nodes in the internal region) (taking one node as the reference phase angle); p is the impedance parameter vector of the erroneous line to be estimated, for example... , , Let p represent the resistance and reactance of the faulty line k, respectively. When there are multiple branches to be corrected, p is a column vector consisting of the impedance parameters of each faulty line arranged sequentially.

[0230] The second step is to construct the augmented state estimation model. This augmented state estimation model is a measurement equation model based on weighted least squares criterion, i.e. , where z is the measurement vector of the simplified network (internal region including boundary nodes), which includes measurements such as node voltage, node injected power, and branch start and end power; To augment state variables Let W be the nonlinear measurement function of the independent variable; W is the measurement weight matrix, which is taken as the inverse of the measurement error covariance matrix. Since the impedance parameter has been extended to a state variable, the measurement function... The partial derivatives with respect to parameter p are included in the Jacobian matrix. Among them.

[0231] The third step is to iteratively solve for the estimated values ​​of the error parameters. The Newton-Gaussian iterative method is used to solve the above weighted least squares model, and the corrected equation for the k-th iteration is: Status updated to ,in In order to be in The augmented Jacobian matrix calculated at point , This is the augmented information matrix (gain matrix). It represents the state correction between two adjacent iterations. The norm is less than the preset convergence criterion (taken as 10). -3 When the iteration terminates, the augmented state variables are... The component corresponding to the impedance parameter is the estimated value of the error parameter. .

[0232] The fourth step, the specific steps for parameter correction and pass rate recovery, are as follows: (a) The impedance parameter estimate obtained by the iteration convergence in the third step is... (a) Replace the original erroneous impedance values ​​of the corresponding erroneous lines in the power grid parameter database; (b) Based on the updated parameter database, re-perform a routine state estimation for the entire network; (c) Calculate the pass rate of the state estimation of the entire network after correction and compare it with that before correction to verify the correction effect.

[0233] The pass rate is calculated as follows: For a certain type of measurement (such as voltage amplitude, active power, or reactive power), the number of measurements for which the absolute value (or normalized residual) of the difference between the estimated value and the measured value after statistical state estimation does not exceed a preset pass threshold is divided by the total number of measurements of that type, and then multiplied by 100% to obtain the pass rate of that type of measurement. The overall pass rate is calculated by dividing the total number of qualified measurements of all types by the total number of measurements, and then multiplying by 100%. That is, pass rate = (number of measurements with residuals not exceeding the preset pass threshold / total number of measurements) × 100%.

[0234] Furthermore, if the pass rate of the overall network state estimation still fails to recover to the expected level after parameter correction (e.g., the overall pass rate is still lower than the preset pass rate target), the following steps are taken: (a) Return to the identification stage and re-check whether there are other parameter error lines that have not yet been identified. That is, repeat the initial screening of abnormal areas, the screening of relative influence factors, and the verification of the deviation in the head-end power flow consistency calculation for areas where the current residuals are still abnormally concentrated. Locate the new parameter error lines and augment their impedance parameters as state variables, then re-perform the zonal augmented state estimation; (b) If no new parameter error lines are found after review, appropriately reduce the sensitivity threshold or continue to expand the buffer layer nodes outward in the aforementioned manner to expand the internal area to include more measurement information that contributes to the parameter estimation, and then re-perform the augmented state estimation; (c) Repeat the above process until the pass rate of the overall network state estimation recovers to the expected level. This ensures that even when the pass rate does not meet expectations, the systematic deviation caused by parameter errors can still be gradually eliminated through iterative correction.

[0235] To verify the effectiveness of the technical solution in the embodiments of the present invention, simulation analysis results are provided below.

[0236] (1) Test system and data settings

[0237] The characteristics of line parameter errors in a ring network were verified using an IEEE 14-bus system, which includes 20 branches and 14 power grid nodes. Measurements were generated by superimposing Gaussian noise onto power flow calculations. Different levels of error were added as error parameters to the actual line parameters, and 100 Monte Carlo simulations were performed.

[0238] The IEEE 118-node standard case was used as a simulation platform to verify the effectiveness and scalability of the method in large-scale complex power systems. Lines 69-75, located in the dense ring network area in the middle of the system, were selected for parameter error settings. This line belongs to a multi-circuit complex ring network structure consisting of nodes 68, 69, 70, 71, 75, and 77.

[0239] (2) Verification of error parameter characteristics

[0240] Concealment Verification: A concealment index H is defined as 1 minus the ratio of the number of measurements identified as abnormal by the detection criteria to the total number of measurements actually affected by parameter errors. The closer the H value is to 1, the stronger the concealment of the parameter errors. Simulation results show that under different line locations and error levels, the concealment indices H1 under the node power balance criterion and H2 under the branch head-end consistency criterion remain at high levels. For example, when there is a 10% error in the resistance parameters, H1 is 0.949 and H2 is 0.899, indicating that approximately 94.9% and 89.9% of the affected measurements, respectively, were not identified by the traditional criteria. Even when the parameter error increases to 1.5 times the true value, the traditional criteria still struggle to detect it effectively.

[0241] Validation of Residual Anomaly Clustering Characteristics: The residual anomaly clustering index (AC) is defined as the number of anomalous branches associated with the topology of the faulty line divided by the total number of anomalous branches whose residual changes exceed a threshold. Simulation results show that, under different line locations and load levels, the AC value ranges from 0.750 to 0.885, with approximately 75% to 89% of the anomalous residual branches showing a topological correlation with the faulty branches. As the load level increases, the AC value rises accordingly, exhibiting a clear positive correlation trend.

[0242] (3) Validation of the identification method

[0243] In the IEEE 118-node case study, the method is compared with the maximum normalized residual method, the traditional residual sensitivity method, and the forward-backward generation power flow tracing method. Parameter error levels are set to 30%, 50%, 70%, and 100%, and 100 Monte Carlo simulations are performed for each error level of each parameter error branch.

[0244] Simulation results show that the method of this invention exhibits the highest identification accuracy at all error levels, reaching over 90% at both 70% and 100% error levels. The identification accuracy of the maximum standardized residual method remains at a lower level of 20% to 30%; the traditional residual sensitivity method is between 40% and 85%.

[0245] (4) Validation of the effect of partition augmentation state estimation

[0246] Three partitioning schemes were designed for comparison: S2-1 is a no-buffer layer scheme with 15 nodes in the internal area and 7 nodes at the boundary; S2-2 is a one-layer buffer scheme with 22 nodes in the internal area and 18 nodes at the boundary; S2-3 is a full-network augmentation scheme with a total of 118 nodes.

[0247] The results show that scheme S2-1 reduces network size by 81.36%, while scheme S2-2 reduces it by 66.10%. S2-2 has mean relative errors of -1.91% and 2.24% for resistance and reactance estimation, respectively, with a computation time of 1.287 seconds, achieving a significant reduction in computation time while maintaining high accuracy. After parameter correction, the voltage amplitude, active power, reactive power, and overall qualification rate are improved to 96.23%, 92.70%, 92.38%, and 93.77%, respectively, fully demonstrating that parameter correction can effectively improve the accuracy of state estimation.

[0248] Robustness verification shows that when the impedance error level increases from 20% to 50%, the corrected pass rate remains above 93%. Under different load levels, the parameter estimation accuracy is higher under heavy load conditions because the branch transmission power is greater under heavy load, resulting in richer parameter information carried in the measured values.

[0249] Figure 9 This is a structural block diagram of a line error parameter identification and estimation device considering the power flow distribution of a transmission ring network, provided as an embodiment of the present invention. This device is used to execute the method provided in any of the above embodiments. See also Figure 9 The device may specifically include: a topology search module, an abnormal area screening module, a suspected branch screening module, an error line location module, and a partition estimation module.

[0250] The topology search module is used to obtain the network topology of the power transmission network and the power grid measurement data. Based on the principle of minimizing the node power imbalance, it performs a hierarchical topology search on the network topology, transforming the ring network into a radial structure, and obtaining the branch hierarchy matrix and the branch head and end node matrix.

[0251] The abnormal area screening module is used to calculate the power residual difference index of each branch based on the power grid measurement data and state estimation results, and to screen the abnormal areas according to the power residual difference index to determine the first-level suspicious areas.

[0252] The suspected branch screening module is used to construct a sensitivity matrix of measurement residuals to line impedance parameters within the first-level suspected area, calculate the relative influence factor of each branch, and screen out the second-level suspected branches.

[0253] The error line location module is used to calculate the head-end power flow consistency deviation of the suspected secondary branch, and construct a joint judgment index in combination with the relative influence factor to locate the error line parameter.

[0254] The partition estimation module is used to partition the entire network based on the residual sensitivity relationship. Local networks strongly correlated with parameter error lines are divided into internal regions, and remote networks weakly correlated are divided into external regions. The external regions are simplified using the extended Ward equivalence method. Augmented state estimation is performed on the simplified network after equivalence to obtain the correction values ​​of the error parameters.

[0255] Optionally, the partition estimation module further includes: a partition criterion submodule, used to set a sensitivity threshold and partition the entire network; a partition optimization submodule, used to perform ring network integrity constraints, boundary buffer layer settings, and observability verification; an equivalence simplification submodule, used to perform equivalence simplification of the external region using the extended Ward equivalence method; and an augmented estimation submodule, used to perform augmented state estimation on the simplified network.

[0256] The apparatus provided in the embodiments of the present invention can execute the methods provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects for executing the methods.

[0257] It is worth noting that in the embodiments of the above-mentioned device, the various units and modules included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be realized; in addition, the specific names of each functional unit are only for easy distinction between each other and are not used to limit the scope of protection of the present invention.

Claims

1. A method for identifying and estimating line error parameters considering the power flow distribution of a transmission ring network, characterized in that, Includes the following steps: Acquire the network topology of the power transmission network and power grid measurement data; Based on the principle of minimizing node power imbalance, a hierarchical topology search is performed on the network topology to determine the loop unblocking point, and the ring network in the network topology is transformed into a radial structure to obtain the branch hierarchy matrix and the branch head and end node matrix. Based on the power grid measurement data and state estimation results, the power residual difference index of each branch is calculated; Based on the power residual difference index, abnormal areas of the transmission network are initially screened to identify first-level suspicious areas; Within the first-level suspected area, a sensitivity matrix based on the measurement residual to the line impedance parameter is constructed, and the relative influence factor of each branch is calculated. Second-level suspected branches are then selected based on the relative influence factor. Calculate the deviation of the power flow consistency estimation at the beginning and end of the suspected secondary branch, and locate the line with incorrect impedance parameters in combination with the relative influence factor. The transmission network is partitioned based on the residual sensitivity relationship. Local networks that are strongly correlated with lines with parameter errors are classified as internal regions, and remote networks that are weakly correlated with lines with parameter errors are classified as external regions. The external region is simplified using the extended Ward equivalence method. Augmented state estimation is then performed on the simplified network after equivalence to obtain the correction values ​​for the error parameters.

2. The method for identifying and estimating line error parameters considering power flow distribution in a transmission ring network according to claim 1, characterized in that, The steps for performing a hierarchical topology search on the network topology based on the principle of minimizing node power imbalance include: Step 1. Form the node-branch association matrix A based on the node and branch connection information; Step 2. Find all nodes with a degree of 1 in the node-branch association matrix A, mark the branches connected to these nodes with a degree of 1 as branches of the same level in the radial network, write them into the branch level matrix L, and remove the nodes and branches from the association matrix A. Step 3. Repeat step 2 until there are no more nodes with a degree of 1; If there are still unmarked nodes at this point, proceed to step 4; Step 4. Calculate the power imbalance of all remaining nodes, and determine the unloop point according to the principle of minimizing the node power imbalance; All branches connected to the unloop point are marked and removed as branches at the same level. Step 5. Repeat step 4 until all nodes are marked and the association matrix A is a zero matrix; The power imbalance at node j is shown below: in, , This represents the imbalance between active and reactive power at node j. , This represents the measured values ​​of active and reactive power injected into node j; This represents the nodes that are directly connected to node j, excluding node j itself. , This represents the measured active and reactive power values ​​of the branch jk directly connected to node j. This represents the complex power imbalance at node j.

3. The method for identifying and estimating line error parameters considering power flow distribution in a transmission ring network according to claim 1, characterized in that, The steps for preliminary screening of abnormal areas in the power transmission network based on the power residual difference index are as follows: Calculate the power residual difference index of each branch. If the power residual difference index of multiple consecutive electrically adjacent branches exceeds the preset threshold, the connected part containing these consecutive abnormal branches will be marked as a first-level suspicious area. The power residual difference indexes for each branch are shown below: in, , This represents the estimated active and reactive power at the beginning of the branch; , This represents the estimated active and reactive power at the end of the branch. The power residual difference index represents the power residual difference of branch ij; , This represents the measured values ​​of active and reactive power in branch ij; , This indicates the measured values ​​of active and reactive power of branch ji.

4. The method for identifying and estimating line error parameters considering power flow distribution in a transmission ring network according to claim 1, characterized in that, The steps for calculating the relative influence factor of each branch include: Based on the linear approximation relationship between the measurement residual and the line impedance parameters, a residual sensitivity matrix is ​​constructed. ,Right now: Where H represents the standard measurement Jacobian matrix; W represents the measurement weight matrix; This represents the matrix of partial derivatives of the measurement function with respect to the parameters; Represents a unit vector; Based on residual sensitivity matrix Calculate the relative impact factor ,Right now: in Indicates relative impact factor; Indicates the region The set of all measurements with non-compliant residuals; This represents the sum of the variation amplitudes of the power measurement residuals of the i-th branch when the parameters of line l change by a unit; Represents the weighted term of the residuals; This represents the maximum value of the influence factor of all possible branches; The relative impact factors are sorted by size, and the branches that rank in the top N are designated as secondary suspicious branches.

5. The method for identifying and estimating line error parameters considering power flow distribution in a transmission ring network according to claim 4, characterized in that, The parameter error branch refers to the branch with the largest joint judgment index; The joint judgment indicators are as follows: In the formula, This represents the relative estimation deviation; , These represent the power values ​​for different branches; , Denotes the weight coefficients, and satisfies ; These are joint judgment indicators.

6. The method for identifying and estimating line error parameters considering power flow distribution in a transmission ring network according to claim 1, characterized in that, The steps for partitioning the entire network based on residual sensitivity relationships include: If the overall sensitivity of a branch is greater than or equal to the sensitivity threshold, then the first and last nodes of the branch are classified into the inner region; if the overall sensitivity of a branch is less than the sensitivity threshold and is directly connected to a node in the inner region, then the node is defined as a boundary node; nodes with an overall sensitivity less than the sensitivity threshold are classified into the outer region. Adjustments were made to the preliminary zoning results, including: ensuring the integrity of the ring network selected in the initial screening of abnormal areas, adding buffer layer nodes outside the boundaries, and verifying the observability of the internal areas.

7. The method for identifying and estimating line error parameters considering power flow distribution in a transmission ring network according to claim 1, characterized in that, The steps for simplifying the external region using the extended Ward isometry method include: The admittance matrix of all network nodes is divided into blocks according to internal regions, boundaries, and external regions; External nodes are eliminated by Gaussian elimination to obtain the equivalent admittance matrix and equivalent injected current at the boundary nodes; Eliminate the external PQ nodes and retain the external PV nodes, and calculate the equivalent admittance between the boundary nodes and the external PV nodes; Virtual PV nodes are set at the boundary nodes to maintain the voltage support characteristics of the external generator; the virtual PV nodes do not provide active power in the ground state, and the voltage amplitude of the virtual PV nodes is the same as that of the corresponding boundary nodes.

8. The method for identifying and estimating line error parameters considering power flow distribution in a transmission ring network according to claim 1, characterized in that, The steps for performing augmented state estimation on the simplified network after equivalence include: The impedance parameters of the faulty line are extended to state variables. ; These are the regular state variables, including the voltage phase angle vector θ and voltage magnitude vector U of each node; p is the impedance parameter vector of the line with parameter errors. Constructing an augmented state estimation model on the simplified network The estimated values ​​of the error parameters are obtained through iterative solution; where z is the measurement vector of the simplified network; To augment state variables is the nonlinear measurement function of the independent variable; W is the measurement weight matrix; Replace the erroneous parameter values ​​in the database with the estimated values, and re-evaluate the normal state of the entire network to restore the pass rate.

9. The method for identifying and estimating line error parameters considering power flow distribution in a transmission ring network according to claim 1, characterized in that, Assume the internal region, including boundary nodes, has a total of There are n nodes, and the number of parameters to be estimated is 1. Number of measurements in the internal area satisfy ,in, The dimension representing the regular state variables; Indicates the number of augmented parameter variables.

10. A device for identifying and estimating line error parameters considering the power flow distribution of a transmission ring network, based on the method described in any one of claims 1-9, characterized in that, include: The topology search module is used to obtain the network topology of the power transmission network and the power grid measurement data. Based on the principle of minimizing the node power imbalance, it performs a hierarchical topology search on the network topology, transforms the ring network into a radial structure, and obtains the branch hierarchy matrix and the branch head and end node matrix. The abnormal area screening module is used to calculate the power residual difference index of each branch based on the power grid measurement data and state estimation results, to screen the abnormal areas and identify first-level suspicious areas. The suspected branch screening module is used to construct a residual sensitivity matrix in the first-level suspected area, calculate the relative influence factor of each branch, and screen out the second-level suspected branches. The error line location module is used to calculate the head-end power flow consistency estimation deviation of the secondary suspected branch, construct a joint judgment index by combining the relative influence factor, and locate the error line parameter. The partition estimation module is used to partition the entire network based on the residual sensitivity relationship, simplify the external region using the extended Ward equivalent method, perform augmented state estimation on the simplified network, and obtain the correction values ​​of the error parameters.

Images