A weighted average fast forward-backward substitution robust state estimation method with a simple loop network

Abstract

The present invention discloses a weighted average fast forward-backward substitution robust state estimation method for a simple loop network. In the present invention, the loop network is decoupled through the node at the location with the minimum power imbalance in the loop network, and the simple loop network system is decoupled into a radial network. The shunt susceptance of the branches in the loop network is equivalently processed to facilitate subsequent power compensation. Then, the weighted average fast forward-backward substitution robust state estimation method is used to perform weighted average estimation on each branch starting from the bottom layer, and the power flow is backward substituted starting from the root node to obtain the system state estimation voltage correction value. Finally, power compensation is performed on the loop network branches where the loop decoupling points are located, so that the power distribution conforms to the actual distribution and further makes the voltages at the loop decoupling points consistent.

Description

Technical Field

[0001] The present invention relates to the field of power system state estimation, and particularly designs a weighted average fast forward-backward substitution robust state estimation method for a simple loop network. Background Technique

[0002] In a loop network, the influence law of bad data on state estimation is more complex, and the phenomena of residual pollution and submergence are more serious. Moreover, in actual situations, most complex networks are loop networks. Studying the robust estimation method for loop networks is of great significance for engineering practical applications. The existing global least squares state estimation method has high estimation accuracy and good robust performance, but the calculation time is long and it is difficult to apply to actual complex systems. In the case of algorithm non-convergence, the global least squares method has poor processing ability for bad data. The existing forward-backward substitution robust state estimation method has high estimation accuracy and can track and identify bad data, further improving the convergence performance of the forward-backward substitution robust state estimation. However, the existing forward-backward substitution method mainly aims at radial networks, making it difficult to apply to systems with loop networks, greatly restricting the application of the forward-backward substitution robust state estimation method. Summary of the Invention

[0003] The purpose of the present invention is to provide a weighted average fast forward-backward substitution robust state estimation method for a simple loop network, so as to solve the problems raised in the above background technique.

[0004] To achieve the above purpose, the present invention provides the following technical solution: A weighted average fast forward-backward substitution robust state estimation method for a simple loop network, including the following steps:

[0005] 1. Obtain the basic data of the power grid. The basic data of the power grid includes the structural parameters of the power grid and various types of measurement information. The measurement information can be divided into seven categories, namely node voltage amplitude, node injected active power, node injected reactive power, branch head active power, branch head reactive power, branch end active power, and branch end reactive power.

[0006] 2. According to the structural parameters and measured quantities of the power grid system, perform topological analysis on the loop network, and according to the power imbalance of the nodes in the loop network, perform loop-opening processing at the location where the power imbalance is the smallest.

[0007] 3. According to the topological structure after loop-opening, perform equivalent processing on the shunt susceptance of the loop network branches.

[0008] 4. Combine the system network after loop-opening, preprocess the basic data of the power grid, and replace the obvious bad data. The preprocessing of bad data is to first identify the measurement data and replace the obvious wrong data before performing state estimation, increasing the robust performance of the algorithm.

[0009] 5. Starting from the underlying branch, perform weighted averaging state estimation on the branches of each layer to obtain the state estimation value of the power grid.

[0010] 6. Starting from the top - layer branch, according to the voltage magnitude of the top - layer head node and the power distribution of the system, perform power flow back - substitution to correct the node voltage and obtain the state estimation value of the power grid after voltage correction.

[0011] 7. Perform power compensation on the branch where the loop - opening point is located to make it close to the actual distribution and obtain the state estimation correction value of the power grid. After the loop - opening process, due to the mismatch between the actual power distribution and the calculated power distribution at the loop - opening point, the voltages at the loop - opening points are inconsistent. Therefore, it is necessary to perform power compensation on the branch where the loop - opening point is located to make the power distribution meet the actual power distribution.

[0012] 8. Correct the weight of the measured quantity according to the state estimation value correction amount.

[0013] 9. Judge whether the difference between the current state estimation correction value and the previous state estimation correction value and the voltage magnitude difference at the loop - opening point are both less than the given convergence criterion. If it converges, end the state estimation; otherwise, return to step 5.

[0014] Preferably, the specific steps in step 2 are as follows: ① According to the structural parameters of the system, form the node - branch incidence matrix AA. ② According to the node - branch incidence matrix AA, obtain the branch numbers and node numbers at the bottom layer currently, and form the network hierarchy matrix LL and the branch start - end node matrix MM with them. ③ Remove the relevant branches of the current layer from the node - branch incidence matrix. ④ Detect whether the node - branch incidence matrix is empty. If it is empty, it means the topological analysis is over and exit the loop; if it is not empty, go to step ⑤. ⑤ According to the node - branch incidence matrix AA, judge whether there is only a loop network (whether the number of nodes with degree 1 is 0). If there is still a radial network, go to step ②; if there is only a loop network, go to step ⑥. ⑥ Record the nodes (with degree greater than 1) in the loop network, and calculate the power imbalance of each node in the loop network according to the formula Calculate the power imbalance of each node in the loop network from the measured data, and take the node A with the smallest power imbalance as the loop - opening point. In the formula, ΔP, ΔQ, and ΔS are the active - power imbalance, reactive - power imbalance, and complex - power imbalance respectively; P i ,Q i are the measured injected active power and injected reactive power of node i respectively; P ik ,Q ikThe active power measurements and reactive power measurements of all branches connected to node i, respectively. ⑦Find all branches in the loop network connected to node A as the branch level matrix LL of the next layer, and the head and end node matrix MM of the relevant branches. ⑧Remove the relevant branches of the current layer from the node-branch incidence matrix. ⑨Check whether the node-branch incidence matrix is empty. If it is empty, it means that the topology analysis is over, and the loop is exited; if it is not empty, go to step ⑤.

[0015] Preferably, the specific steps in step 3 are as follows: To facilitate subsequent power compensation for the branch where the loop opening point is located, it is first necessary to perform equivalent processing on the branches in the loop network, that is, equivalent the shunt susceptance in the loop network branches to a reactive power source, and add it to the injected reactive power measurement data of each node in the loop network and the reactive power measurement data of each branch, as Figure 1 . The specific equivalent method is as follows: A. Equivalent the shunt susceptance to the injected reactive power of each node in the loop network, as shown in the following formula In the formula, Q i , Q j , Q k are the injected reactive power measurement data of each node before equivalence; Q′ i , Q′ j , Q′ k are the injected reactive power measurement data of each node after equivalence. B. Equivalent the shunt susceptance to the reactive power flow of each branch in the loop network, see formula

[0016] In the formula: Q ij and Q ji , Q jk and Q kj , Q ki and Q ik are the head and end reactive power measurement data of each branch in the loop network before equivalence; Q ij ′ and Q ji ′, Q jk ′ and Q kj ′, Q ki ′ and Q ik ′ are the head and end reactive power measurement data of each branch in the loop network after equivalence.

[0017] Preferably, the specific steps in step 4 are as follows: A. According to the formula judge the power balance of the branch end node. If it is not balanced, it means that there is a bad data at the branch end node, and the node power estimation is not performed on this node, and jump to D. Where k h ∈j represents the nodes directly connected to node j, but does not include node j itself; P j , Q jrespectively represent the active power injection and reactive power injection of node j; ε1 represents the upper limit threshold of the balance judgment determined according to the measurement type and its reference value. B. Perform node power estimation on this node, and obtain the estimated value at the head end deduced from the measured values at the tail end through the power flow calculation method. The node power estimation process is as follows: B1. Establish the node power balance equation

[0018] where: k = k1, k2,... k l , i; k ∈ j represents all the nodes adjacent to node j, but not including node j itself; w pjk , w pj , w qjk , w qj respectively represent the weights corresponding to the measurement data of four types: active power of the branch, injected active power, reactive power of the branch, and injected reactive power; P jk , P j , Q jk , Q j represent the estimated values after estimation; P mjk , P mj , Q mjk , Q mj represent the measured values corresponding to the estimated values. B2. Solve the node power balance equation In the formula: is the estimated value after estimation. C. According to Equation judge whether the power and voltage amplitude at the head and tail ends are consistent. If not, it means there are bad data in the head-end measurement, and replace it with the estimated value deduced from the tail end, then jump to the single-branch state estimation; if they are consistent, it means there are no bad data at both the head and tail ends, and directly perform the single-branch state estimation. In the formula: and P ij respectively represent the active power deduced from the tail end of the branch to the head end of the branch, and the measured value of the active power at the head end of the branch; and Q ij respectively represent the reactive power deduced from the tail end of the branch to the head end of the branch, and the measured value of the reactive power at the head end of the branch; ε2 represents the upper limit threshold of the balance judgment determined according to the power measurement type and its reference value. The method for determining the upper limit threshold is as follows: In the formula: ε is the upper limit threshold of different measurement types; E is the allowable estimation error rate of different measurement types, where it is required that the active power is not higher than 2%, the reactive power is not higher than 3%, the voltage of 220 kV and above is not higher than 0.5%, and the voltage below 220 kV is not higher than 2%. Z b is the system reference value of voltage or power, and Z base is the reference value of the estimation error rate determined from the system reference value of voltage or power. D. Obtain the estimated value at the head end deduced from the measured values at the tail end through the power flow calculation method. E. According to Equation Judge whether the power at the head and the end is consistent. If it is consistent, and there is no bad data or the bad data has been replaced in the lower branches of the end node, it indicates that there is an error in the injection power of the end node. After replacing it, perform single-branch state estimation.

[0019] Preferably, the specific steps of the weighted averaging state estimation in step 5 are as follows:

[0020] A. According to the measurement data at the end of the branch, use the power flow method formula to calculate the calculated value at the head of the branch. In the formula is the measured value or the previous estimated value of the complex power at the end of the branch, is the measured value or the previous estimated value of the voltage at the end of the branch, Z ij is the branch impedance, are the calculated complex voltage and complex power at the head of the branch respectively.

[0021] B. According to the calculated value at the head obtained from the end and the measured value at the head, perform weighted averaging according to the formula to obtain the estimated value at the head of the branch. In the formula, h f is the estimated value at this point, h mf is the measured value at the head, is the calculated value at the head deduced from the end, w f is the weight of the measurement at the head, w t is the weight of the measurement at the end.

[0022] Preferably, the specific process of the power compensation in step 7 is as follows: As shown in the appendix Figure 2 , the power imbalance at the loop-opening points B41 and B42 is: In the formula: ΔU represents the voltage difference at the loop-opening node; Z Σ represents the impedances of all branches in the loop network: To make the power flow at the loop-opening point close to the actual power flow, the power imbalance can be reversely superimposed between the two nodes, which is the power compensation. And to prevent oscillation during the iteration process from causing non-convergence, an oscillation adjustment coefficient α is added before the power compensation: In the formula: α < 1.0 is the adjustment coefficient, which can avoid oscillation during the iteration process.

[0023] Preferably, the weights in step 8 are corrected, and the model of the exponential weight function is: In the formula is the updated weight, is the initial fixed weight, r Nis the standardized residual. Before correcting the weights according to the residuals, it is first necessary to classify and standardize the residuals of the measured values. The specific process is as follows: The residual dz is divided into three types according to the measurement type: the voltage-type measurement residual dz_V, the active power-type measurement residual dz_P, and the reactive power-type measurement residual dz_Q. Assuming that the measurement residuals of the three types all follow a normal distribution, divide the above three residuals by the corresponding standard deviations to obtain the standardized residual r N , and then according to the standardized residual r N , use an exponential weight function to obtain the correction weight of each measured value.

[0024] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0025] The present invention unties the loop of the loop network through the node at the location with the smallest power imbalance in the loop network, unties the simple loop network system into a radial network, and equalizes the shunt susceptance of the branches in the loop network to facilitate subsequent power compensation. Then, the weighted average fast forward-backward substitution robust state estimation method is used to perform weighted averaging estimation on each branch from the bottom layer, and the power flow is then back-substituted from the root node to obtain the system state estimation voltage correction value. Finally, by performing power compensation on the loop network branches where the loop is untied, the power distribution is made to conform to the actual distribution, thereby making the voltages at the loop-untying points consistent. The present invention extends the forward-backward substitution state estimation method that is only applicable to radial networks to systems containing simple loop networks, and retains a certain accuracy and a relatively fast calculation speed. Moreover, as the scale of the system continues to increase, the advantage of this algorithm in terms of calculation efficiency will become more obvious. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 Schematic diagram of the equivalent shunt susceptance of loop network branches

[0027] Figure 2 Schematic diagram of power compensation for the branch at the loop-untying point

[0028] Figure 3 Simple loop network 14-node system

[0029] Figure 4 Schematic diagram of node power balance

[0030] Figure 5 Schematic diagram of weighted averaging state estimation

[0031] Figure 6 Flowchart of weighted average fast state estimation for a system containing a simple loop network DETAILED DESCRIPTION OF THE INVENTION

[0032] In order to verify the correctness of the present invention, a simulation analysis is carried out in combination with the IEEE 14-node system containing a simple loop network. The specific steps are as follows:

[0033] 1. Establish a simulation system, as Figure 3 shown, which is the IEEE 14-node system with a simple loop network. The construction method of this system is to remove the branches 2-4, 2-5, 4-9, 6-13, 10-11, and 13-14 from the IEEE 14 standard system to ensure that there is only a simple loop network and a radial network in the system. The base voltage of the system is 220 kV, and the base capacity is 100 MVA. The measurement data is obtained by superimposing a normal error on the true value of the power flow calculation; 3-5 times the threshold is randomly superimposed on the true value of the power flow calculation to generate bad data; the oscillation coefficient of power compensation is set to 0.5.

[0034] According to the threshold calculation method mentioned above, the power base value corresponding to the 220 kV voltage level is 305 MVA, and the voltage base value is 264 kV. From this, the thresholds of the active power, reactive power, and voltage amplitude of the system can be calculated as: 0.0610, 0.0915, and 0.0528 respectively.

[0035]

[0036]

[0037]

[0038] 2. Conduct a topological structure analysis of the system, and based on the measured values, find the node with the smallest power imbalance in the loop network and perform loop-opening processing on the loop network.

[0039] Based on the measured values, calculate the unbalanced power of the nodes in the loop network, and the calculation results are shown in Table 1.

[0040] Table 1 Unbalanced power of loop network nodes

[0041]

[0042] The node number with the smallest power imbalance is 1. Therefore, taking node 1 as the loop-opening point, perform loop-opening on the loop network, and conduct a topological analysis of the system to obtain the network hierarchy matrix LL and the branch start and end node matrix MM of the system.

[0043]

[0044]

[0045] 3. According to the topological structure after loop-opening, perform equivalent processing on the shunt susceptance of the loop network branches.

[0046] The specific equivalent method is as follows:

[0047] ①Equivalent the shunt susceptance to the reactive power injection at each node in the loop network, as shown in Equation (4).

[0048]

[0049] Where Q i , Q j , Q k are the measured data of reactive power injection at each node before equivalence; Q′ i , Q′ j , Q′ k are the measured data of reactive power injection at each node after equivalence respectively.

[0050] ②Equivalent the shunt susceptance to the reactive power flow in each branch of the loop network, as shown in Equation (5).

[0051]

[0052] Where: Q ij and Q ji , Q jk and Q kj , Q ki and Q ik are the measured data of the initial and end reactive power at each branch in the loop network before equivalence; Q ij ′ and Q ji ′, Q jk ′ and Q kj ′, Q ki ′ and Q ik ′ are the measured data of the initial and end reactive power at each branch in the loop network after equivalence respectively.

[0053] The measured data after equivalence are shown in Table 2.

[0054] Table 2 Measured values after equivalent shunt susceptance

[0055]

[0056] 4. Combine the system network after loop opening and preprocess the basic data of the power grid to replace the obvious incorrect bad data.

[0057] According to the network hierarchical structure matrix LL, starting from the bottom - layer branches, perform node power estimation and bad data identification for each branch. The system state variables after bad data identification are shown in Table 3.

[0058] The specific steps of node power estimation are as follows:

[0059] According to Kirchhoff's law, the injection power and outflow power of any node in the power system should be balanced.

[0060] For exampleFigure 4 As shown, there should be Among them represents the power injected into node j, and k ∈ j represents all nodes adjacent to node j, excluding node j itself.

[0061] Since there are certain errors in both the measured values and the estimated values after branch state estimation, according to the principle of WLS, the measured or estimated values of node power can be subjected to state estimation to satisfy Kirchhoff's law, facilitating the participation in the state estimation calculation of the upper-level branch.

[0062]

[0063]

[0064] Among them: k = k1, k2,... k l , i; k ∈ j represents all nodes adjacent to node j, excluding node j itself; w pjk , w pj , w qjk , w qj respectively represent the weights corresponding to the measured data of four types: branch active power, injected active power, branch reactive power, and injected reactive power; P jk , P j , Q jk , Q j represent the estimated values after estimation; P mjk , P mj , Q mjk , Q mj represent the measured values corresponding to the estimated values.

[0065] Solving the above equation gives:

[0066]

[0067]

[0068] In the formula: is the estimated value after estimation.

[0069] 5. According to the LL and MM matrices, starting from the end nodes of the bottom-layer branches 9, 11, 7, 13, and 12, perform weighted average state estimation on each branch until the top-layer branch 0 is reached, obtaining the initial values of the system state variables. At this time, the system state variables are shown in Table 3.

[0070] The specific steps of weighted average state estimation ( Figure 5 ) are as follows:

[0071] For a simple branch, assuming it is a fully measured system, that is, there are measured quantities: Vim , V jm , P im , P jm , Q im , Q jm , P ijm , Q ijm , P jim , Q jim 。

[0072] From the end of the branch, measure V at the end jm , P jim , Q jim Based on this, the calculated values of the branch head can be calculated according to the following formula according to the forward power flow calculation method ic , P ijc , Q ijc 。

[0073]

[0074] In the formula is the measured value or the previous estimated value of the complex power at the end of the branch, is the measured value or the previous estimated value of the voltage at the end of the branch, Z ij is the branch impedance, are the calculated complex voltage and complex power at the branch head respectively.

[0075] Each time, the calculated value is weighted-averaged with the initial measured value at the branch head according to Equation (11) to obtain a new estimated value, which is used in the calculation of the next-level branch.

[0076]

[0077] In the formula, h f is the estimated value at this point, h mf is the measured value at the head, is the estimated value of the head deduced from the end, w f is the weight of the head measurement, w t is the weight of the end measurement.

[0078] 6. According to the LL and MM matrices, starting from the head node of the top-level branch, correct the voltage amplitudes of each node with the head voltage and the overall power distribution of the system to obtain the voltage correction value of the system state quantity. The specific results are shown in Table 3.

[0079] Table 3 Node voltage amplitude changes

[0080]

[0081]

[0082] 7. Perform power compensation on the branch where the loop-opening point is located.

[0083] Since the loop network is opened at the loop-opening point, there is a certain difference between the actual power distribution and the calculated result, which leads to inconsistent voltage amplitudes at the loop-opening point. Therefore, it is necessary to perform power compensation on the branch where the loop-opening point is located. After power compensation, the voltage amplitudes at the loop-opening point each time are shown in Table 4.

[0084] The specific steps of power compensation are as follows:

[0085] For example Figure 2 , the power imbalance at loop-opening points B41 and B42 is:

[0086]

[0087]

[0088] In the formula: ΔU represents the voltage difference at the loop-opening node; Z Σ represents the impedance of all branches in the loop network:

[0089]

[0090]

[0091] To make the power flow at the loop-opening point close to the actual power flow, the power imbalance can be reversely superimposed between two nodes, which is power compensation. And to prevent oscillation during the iteration process from causing non-convergence, an oscillation adjustment coefficient α is added before power compensation:

[0092]

[0093]

[0094] In the formula: α < 1.0 is the adjustment coefficient, which can avoid oscillation during the iteration process.

[0095] Table 4 Variation of voltage amplitudes at loop-opening points during iteration

[0096]

[0097]

[0098] 8. According to the corrected values of the system state variables, classify and standardize them according to three types: voltage amplitude, active power, and reactive power, and correct the weights of each type through an exponential weight function.

[0099] An exponential weight function is used to correct the weights of the measured quantities. The model of the exponential weight function is as follows:

[0100]

[0101] In the formula is the updated weight, is the initial fixed weight, r N is the standardized residual.

[0102] Before correcting the weight according to the residual, it is first necessary to classify and standardize the residual of the measured quantity. The specific process is as follows:

[0103] The residual dz is divided into three types according to the measurement type: the voltage type measurement residual dz_V, the active power type measurement residual dz_P, and the reactive power type measurement residual dz_Q. Assuming that the measurement residuals of the three types all satisfy the normal distribution, the above three residuals are divided by the standard deviation of the corresponding type to obtain the standardized residual r N , and then according to the standardized residual r N , an exponential weight function is used to obtain the correction weight of each measured value.

[0104] 9. Judge whether the difference in the voltage amplitude of the loop-opening point between the state estimation correction value this time and the state estimation correction value last time is less than the given convergence criterion. If it converges, end the state estimation; otherwise, return to step 5.

[0105] 10. Index evaluation

[0106] 10.1 Definition of evaluation index

[0107] In order to evaluate the accuracy and robustness of the method, the average error and the maximum error of the system are defined.

[0108]

[0109]

[0110] In the formula: S1 is the average error of the system; S2 is the maximum error of the system; and x i are the estimated value and the true value of the i-th state variable respectively.

[0111] 10.2 Simulation scheme

[0112] 1. When no bad data is set, based on the true value of the power flow calculation, only measurement errors are superimposed according to formula (21), and 100 independent experiments are carried out on the traditional WLS method, the complete forward-backward substitution robust state estimation method and the method of the present invention respectively, and the evaluation indexes are calculated.

[0113] z = PF_T(1 + N(0, std_error)) (21)

[0114] Wherein, PF_T is the power flow calculation value; std_error is the standard deviation of the measurement error. According to different measurement types, the standard deviations of the voltage amplitude, the injected power, and the branch power are set to 0.004, 0.01, and 0.008 respectively.

[0115] 2. In the backward / forward sweep robust method for the radial network, the robust performance of the weighted average fast backward / forward sweep robust state estimation method has been proven. Therefore, only for the simple loop network, single measurement error, multiple measurement errors at the same node, and multiple measurement errors in the associated branches are respectively set to evaluate the method of the present invention.

[0116] Table 5 Various simulation methods

[0117]

[0118] 10.3 Simulation results and analysis

[0119] 10.3.1 When no bad data is set, the simulation results of each method are as follows:

[0120] Table 6 Simulation results of normal measurement

[0121]

[0122] From the perspective of the accuracy of the results, the global least squares method has the highest accuracy, followed by the backward / forward sweep method with the least squares method for each branch. The estimation accuracy of the method of the present invention is slightly lower than that of the backward / forward sweep method with the least squares method for each branch. The estimation accuracy of the global least squares method is higher than that of the backward / forward sweep method. The main reasons are as follows:

[0123] 1. During the process of equivalent shunt susceptance to ground, the voltages of each loop network node are not real values, which in turn leads to certain errors in the injected reactive power of the nodes and the reactive power of the branches after equivalence.

[0124] 2. When performing power compensation, a regulation coefficient is added to prevent the calculation results from oscillating and not converging. Different regulation coefficients have a certain impact on power compensation.

[0125] 3. The process of power compensation sometimes conflicts with the goal of minimizing the residual. However, in order to make the power distribution conform to the actual power flow, the residual will be increased to a certain extent.

[0126] From the perspective of calculation speed, the calculation speed of the method of the present invention is much faster than the forward and backward method in which each branch adopts the least squares method. This is because the algorithm of the present invention adopts weighted averaging calculation. Compared with the forward and backward method in which each branch adopts the least squares method, the number of calculations for each branch is greatly reduced. In this example, the calculation speed of the method of the present invention is slightly lower than that of the global least squares method. As the scale of the system becomes increasingly complex, the calculation time required for the global least squares method will increase significantly, and the advantage of the method of the present invention in calculation efficiency will become more obvious. 10.3.2 When special bad data is set, the simulation results of each method are as follows:

[0127] Table 7 Simulation results of defective data at special locations

[0128]

[0129]

[0130] After setting the bad data of the special position, it can be seen that when setting the single bad power data, the method of the present invention and the global least squares method both have high estimation accuracy. At the same time, due to the addition of bad data preprocessing, the estimation accuracy of the method of the present invention is higher than that of the global least squares method. When setting the single bad voltage data, the estimation accuracy of the several methods is greatly reduced. The estimation accuracy of the method of the present invention is 10 -4 The order of magnitude is one order of magnitude lower than that of the global least squares method, because the bad voltage data in the ring network has a greater impact on the voltage at the de-ring point, which will cause a certain deviation in power compensation. When there are more associated bad data, the mutual influence between the bad data in the ring network is closer, resulting in a decrease in the calculation accuracy of several methods. Overall, the global least squares method has the highest estimation accuracy, and each branch uses the forward and backward method of the least squares method. Secondly, the estimation accuracy of the method of the present invention is slightly lower than that of the forward and backward method of the least squares method in each branch.

[0131] In terms of calculation speed, the calculation speed of the method of the present invention is significantly faster than the forward and backward method in which each branch adopts the least squares method. When there is less bad data, the calculation speed of the method of the present invention is faster than the global least squares method. As the number of bad data increases, the number of iterations of the method of the present invention increases significantly, resulting in a decrease in speed. As the scale of the system becomes increasingly complex and larger, the calculation time of the global least squares method and the forward and backward method in which each branch adopts the least squares method will increase significantly. However, since the method of the present invention does not require the least squares state estimation, the advantage in calculation speed is more obvious.

Claims

1. A weighted average fast forward-backward substitution robust state estimation method with a simple loop network, characterized in that It includes the following steps: (1) Obtain the basic data of the power grid. The basic data of the power grid includes the structural parameters of the power grid and various types of measurement information. The measurement information is divided into seven categories, namely the node voltage amplitude, the active power injected into the node, the reactive power injected into the node, the active power at the beginning of the branch, the reactive power at the beginning of the branch, the active power at the end of the branch, and the reactive power at the end of the branch; (2) According to the structural parameters of the power grid and the measurement data, conduct a topological analysis of the loop network, and according to the power imbalance of the loop network nodes, perform loop disconnection at the location with the minimum power imbalance; (3) According to the topological structure after loop disconnection, perform equivalent processing on the shunt susceptance of the loop network branches; (4) Combine the system network after loop disconnection, preprocess the basic data of the power grid, and replace the obviously incorrect bad data; (5) Starting from the bottom-layer branches, perform weighted average state estimation on each layer of branches to obtain the state estimation value of the power grid; (6) Starting from the top-layer branches, according to the voltage amplitude of the top-layer head node and the power distribution of the system, perform power flow back substitution to correct the node voltage and obtain the state estimation value of the power grid after voltage correction; (7) Perform power compensation on the branch where the loop disconnection point is located to make it close to the actual distribution and obtain the corrected state estimation value of the power grid; (8) Update the weights of the measured quantities according to the state estimation value correction amount; (9) Judge whether the difference between the current state estimation corrected value and the previous state estimation corrected value and the voltage amplitude of the loop disconnection point is less than the given convergence criterion; If it converges, end the state estimation; otherwise, return to step (5).

2. A weighted average fast forward-backward substitution robust state estimation method with a simple loop network according to claim 1, characterized in that, The specific steps in step (2) are as follows: ①. According to the structural parameters of the system, form the node-branch incidence matrix AA; ②. According to the node-branch incidence matrix AA, obtain the branch numbers and node numbers of the currently lowest layer, and form the network hierarchy matrix LL and the branch start-end node matrix MM with them; ③ Remove the relevant branches of the current layer from the node-branch incidence matrix; ④ Detect whether the node-branch incidence matrix is empty. If it is empty, it means the topological analysis is over and exit the loop; If it is not empty, perform step ⑤; ⑤ According to the node-branch incidence matrix AA, judge whether there is only a loop network, that is, judge whether the number of nodes with degree 1 is 0. If there is still a radial network, perform step ②; If there is only a loop network, perform step ⑥; ⑥ Record the nodes in the loop network, that is, record the power imbalance with degree greater than 1 nodes, according to Equation calculate each node in the ring network from the measurement data and take the node A with the minimum power imbalance as the loop disconnection point; where ΔP, ΔQ, and ΔS are the active power imbalance, reactive power imbalance, and complex power imbalance respectively; P i , Q i are the measured active power injection and measured reactive power injection at node i respectively; P ik , Q ik are the measured active power and measured reactive power of all branches connected to node i respectively; ⑦ Find all the branches in the loop network connected to node A as the next-layer branch hierarchy matrix LL and the head and end node matrix MM of the relevant branches; ⑧ Remove the relevant branches of the current layer from the node-branch incidence matrix; ⑨ Detect whether the node-branch incidence matrix is empty. If it is empty, it means the topological analysis is over and exit the loop; If it is not empty, perform step ⑤.

3. A weighted average fast forward-backward substitution robust state estimation method with a simple loop network according to claim 1, characterized in that, The specific steps in step (3) are as follows: To facilitate subsequent power compensation for the branch where the loop disconnection point is located, first, it is necessary to perform equivalent processing on the branches in the loop network, that is, equivalent the shunt susceptance in the loop network branches to a reactive power source and add it to the reactive power measurement data injected into each node in the loop network and the reactive power measurement data of each branch.

4. A weighted average fast forward-backward substitution robust state estimation method with a simple loop network according to claim 1, characterized in that, The specific steps in the step (4) are as follows: A. According to the formula judge the power balance of the end node of the branch. If it is unbalanced, it means that there is bad data at the end node of the branch. Do not perform node power estimation on this node and jump to D; where represents the node directly connected to node j, but does not include node j itself; respectively represent the active power injection and reactive power injection of node j; represents the upper limit threshold of the balance judgment determined according to the measurement type and its reference value; B. Perform node power estimation on this node, and obtain the estimated value at the head end deduced from the measured value at the end through the power flow calculation method; C. According to the formula judge whether the power at the head and tail ends is consistent with the voltage amplitude. If it is not consistent, it means that there are bad data in the head-end measurement, and replace it with the estimated value deduced from the tail end, then jump to the single-branch state estimation; if it is consistent, it means that there are no bad data at both the head and tail ends, and directly perform the single-branch state estimation; where: and P ij are the active power obtained by inferring from the branch end to the branch head and the measured value of the active power at the branch head, respectively; and Q ij are the reactive power obtained by inferring from the branch end to the branch head and the measured value of the reactive power at the branch head, respectively; ε2 represents the upper limit threshold of the balance judgment determined according to the power measurement type and its reference value; D. The estimated value of the head end obtained by inferring from the measured value at the end through the power flow calculation method; E. According to the formula judge whether the power at the head and end is consistent. If it is consistent and there is no bad data or the data has been replaced in the lower branches of the end node, it indicates that there is an error in the injection power of the end node. After replacing it, perform single-branch state estimation.

5. A weighted average fast forward-backward substitution robust state estimation method with a simple loop network according to claim 1, characterized in that The specific steps of the weighted averaging state estimation in step (5) are as follows: A. Based on the measurement data at the end of the branch, through the power flow method calculate the calculated values at the beginning of the branch; where is the measured value or the previous estimated value of the complex power at the end of the branch, is the measured value or the previous estimated value of the voltage at the end of the branch, Z ij is the branch impedance, are the calculated complex voltage and complex power at the beginning of the branch respectively; B. According to the calculated head-end value obtained from the end and the measured value at the head-end, based on Equation perform weighted averaging to obtain the estimated value of the branch head-end; where h f is the estimated value of the branch head-end, h mf is the measured value at the head-end, is the calculated value of the head-end deduced from the end, w f is the weight of the head-end measurement, w t is the weight of the end measurement.

6. A weighted average fast forward-backward substitution robust state estimation method with a simple loop network according to claim 1, characterized in that, After the loop-opening treatment in step (7), since the actual power distribution at the loop-opening point does not match the calculated power distribution, the voltage at the loop-opening point is inconsistent. Therefore, it is necessary to perform power compensation on the branch where the loop-opening point is located so that the power distribution meets the actual power distribution.

7. A weighted average fast forward substitution robust state estimation method with a simple loop network according to claim 1, characterized in that In the step (8), an exponential weight function is used to correct the weight of the measurement, and the model of the exponential weight function is as follows: In the formula is the updated weight, is the initial fixed weight, and rN is the standardized residual.

Images