Power distribution network line parameter on-line identification method based on PMU measurement

Abstract

This application relates to the field of distribution network line parameter identification technology, specifically to an online identification method for distribution network line parameters based on PMU measurements. This method acquires voltage and current data at both ends of the line using a PMU, extracts positive-sequence phasors, establishes a linear observation equation for a lumped parameter model, and constructs a global observation matrix within a sliding window. Huber estimation is used to construct the objective function, and the threshold and weights are adaptively updated using the median absolute deviation, iteratively solving for line resistance and reactance until convergence. Topology changes are monitored in real time, and the window is reset to dynamically update the observation data. After error verification, the identification results are output. This method combines robust estimation and dynamic adjustment mechanisms to effectively cope with measurement noise and distribution network topology changes, improving the real-time performance and accuracy of parameter identification.

Description

Technical Field

[0001] This application relates to the field of power distribution network line parameter identification technology, specifically to an online identification method for power distribution network line parameters based on PMU measurements. Background Technology

[0002] Distribution network line parameters are core foundational data for power grid operation analysis and control decisions. Their accuracy directly affects the precision of distribution network state estimation, fault location accuracy, and power flow calculation reliability. Currently, distribution network line parameter identification mainly relies on offline measurements or traditional least squares estimation methods, which have many drawbacks in engineering applications.

[0003] Existing methods for identifying distribution network line parameters face significant challenges: offline measurements require power outages, impacting power supply reliability and struggling to adapt to dynamically changing distribution network topologies, leading to delayed parameter updates; traditional least-squares estimation is highly sensitive to measurement noise, model errors, and unknown disturbances, and abnormal data significantly increases identification bias and robustness when the system experiences load fluctuations, distributed power source integration, or line aging; while existing online methods are mostly designed for transmission networks and struggle to effectively address the complex impedances, numerous branches, and frequent disturbances in distribution networks, limiting their engineering applicability. These issues collectively constrain the real-time performance, accuracy, and reliability of parameter identification. Summary of the Invention

[0004] In view of the above, it is necessary to provide an online identification method for distribution network line parameters based on PMU measurements to solve the above problems.

[0005] One embodiment of this application provides an online identification method for distribution network line parameters based on PMU measurements, the method comprising: Acquire voltage and current PMU measurement data at the beginning and end of the distribution network to be identified, and extract positive sequence voltage phasors and positive sequence current phasors; Based on the lumped parameter model of the line, a complex equation for the difference between the positive sequence voltage phasors and the positive sequence current phasors at both ends of the line is established and expanded into a linear observation equation in real form. The observation data of multiple consecutive sampling points within a preset sliding window are vertically spliced ​​together to construct a global observation matrix and a global observation vector. Based on Huber estimation, a parameter identification objective function is constructed. The Huber threshold and weight matrix of the residuals are adaptively updated using the median absolute deviation algorithm. The line resistance and reactance are iteratively solved until the convergence condition is met. The distribution network topology is monitored in real time. When the topology changes, the sliding window is reset. After the newly collected measurement data reaches the window length, the global observation matrix and global observation vector are reconstructed for identification. The identified line resistance and reactance are checked for error and stability, and the final parameter identification results are output.

[0006] Preferably, the complex equations for establishing the positive-sequence voltage phasor difference and positive-sequence current phasor at both ends of the line and expanding them into linear observation equations in real form are as follows: In the formula, Represents the global observation vector; The parameter vector to be identified; This is the model error vector; H represents the measurement noise and unknown disturbance vector; H represents the global observation matrix.

[0007] Preferably, the process of obtaining the global observation matrix and the global observation vector is as follows: Calculate the positive sequence voltage phasor difference between the two ends of the line. ,in , These are the extracted positive sequence voltage phasors at the beginning and end of the line, respectively. Establish complex equations: ,in For positive sequence current phasors and To identify the line resistance and reactance, The imaginary unit; The positive sequence voltage phasor difference and positive sequence current phasor are expanded into real and imaginary parts. Based on the principle that the real parts are equal to the real parts and the imaginary parts are equal to the imaginary parts, a single observation vector and a single observation matrix for a single sampling are constructed. The global observation vector is generated by vertically concatenating the single observation vectors corresponding to all sampling points within the preset sliding window, and the global observation matrix is ​​generated by vertically concatenating the single observation matrices corresponding to all sampling points.

[0008] Preferably, the parameter identification objective function is obtained by summing the Huber loss functions corresponding to the residuals of all observations.

[0009] Preferably, the Huber loss function corresponding to the residual is as follows: If the absolute value of the residual is less than or equal to the Huber threshold, then the value of the Huber loss function is half the square of the residual; If the absolute value of the residual is greater than the Huber threshold, then the value of the Huber loss function is the product of the Huber threshold and the absolute value of the residual, minus half the square of the Huber threshold.

[0010] Preferably, the process of adaptively updating the Huber threshold and weight matrix of the residuals using the median absolute deviation algorithm specifically involves: Based on the residual sequence of the current iteration, the robust estimate of the standard deviation of the residual is obtained by calculating the median absolute deviation; The robust estimation standard deviation is multiplied by a preset confidence coefficient to output the updated Huber threshold, and then truncation constraints are applied to it using preset upper and lower threshold limits. The ratio of the derivative of the Huber loss function to the residual is used as the weight of the corresponding observation to construct the updated weight matrix; when the residual is zero, the corresponding weight is directly output as 1 to prevent division by zero error.

[0011] Preferably, the iterative solution of the line resistance and reactance specifically involves: The line's rated resistance and rated reactance are used as the initial values ​​for the parameter vector to be identified; Based on the weight matrix, global observation matrix, and global observation vector of the current iteration, the updated parameter vector to be identified is calculated using the weighted least squares formula. Calculate the change in the parameter vector to be identified before and after this iteration update, and perform a two-level convergence determination; The iteration stops when the convergence condition is met or the maximum number of iterations is reached, and the current vector of parameters to be identified is output as the preliminary identification result of the line resistance and reactance.

[0012] Preferably, the two-stage convergence determination process is as follows: The initial convergence threshold is set as the first convergence threshold; if the absolute value of the relative difference between the change in the current iteration parameter and the change in the previous iteration parameter is greater than the preset fluctuation ratio, the convergence threshold is tightened from the first convergence threshold to a smaller second convergence threshold, and the maximum number of iterations is increased simultaneously. Accordingly, the convergence condition specifically includes any one of the following: the change in the parameter vector to be identified decreases continuously for a preset number of consecutive times and is less than the current convergence threshold; the number of iterations reaches the current maximum number of iterations limit.

[0013] Preferably, the determination of a change in the topology is based on: obtaining line switching event information or distributed power source access / exit event information in real time through the distribution network dispatching system; when the event information is received, it is determined that the distribution network topology has changed.

[0014] Preferably, the specific process of performing error and stability checks on the identified line resistance and reactance, and outputting the final parameter identification result, is as follows: The relative error between the parameter identification results and the line rated parameters is calculated, as well as the ratio of the range of the parameter identification results over multiple consecutive sliding window periods to their mean. If the relative error does not exceed a preset error threshold and the ratio does not exceed a preset stability threshold, the verification is deemed successful, and the current parameter identification result is output as the final result. If the verification fails, then based on the parameter error and convergence status, perform at least one of the following operations: increase the Huber threshold or increase the maximum number of iterations, and re-execute the parameter identification based on Huber estimation. If the number of iterations for parameter identification reaches the preset maximum limit and the verification still fails, the adjustment will be terminated, a parameter anomaly alarm will be output, and the identification result of the previous convergence cycle will be used as the final result.

[0015] This application has at least the following beneficial effects: To address the shortcomings of traditional methods, which are susceptible to measurement noise, model errors, and unknown disturbances, this paper innovatively uses Huber estimation to replace least squares estimation. By differentially processing the residuals and assigning low weights to outliers, and by dynamically updating the threshold and weight matrix using the median absolute deviation algorithm, the paper effectively suppresses the bias caused by data anomalies and model uncertainty.

[0016] Overcoming the limitations of traditional methods with lagging parameter updates, a real-time topology monitoring and parameter linkage correction mechanism is constructed. When the distribution network topology changes due to fault reconstruction, equipment switching, etc., the system can instantly detect and reset the sliding data window. After new data accumulates to a preset threshold, the observation matrix is ​​quickly reconstructed, achieving dynamic parameter updates. This mechanism ensures that the identification results always match the current network structure, solving the problem of parameter failure caused by topology changes and providing a reliable basis for real-time analysis and control of the distribution network.

[0017] Based on PMU measurement data, this system enables real-time online identification without power outages, avoiding interference with the normal power supply to the distribution network and significantly improving operational flexibility. The algorithm design balances efficiency and accuracy, with moderate computational complexity, allowing for seamless integration into existing distribution network dispatch automation systems. It has low hardware resource requirements and is easy to deploy at scale. The high-precision parameter identification results provide a data foundation for advanced applications such as distribution network state estimation, precise fault location, and power flow optimization, achieving a balance between stability, timeliness, accuracy, and practicality in parameter identification. This provides solid technical support for the efficient operation and intelligent management of the distribution network. Attached Figure Description

[0018] Figure 1 A flowchart of the online identification method for distribution network line parameters based on PMU measurement provided in this application; Figure 2 A diagram illustrating the iterative process for parameter identification provided in this application. Detailed Implementation

[0019] This application proposes an online identification method for distribution network line parameters based on PMU measurements, which is applied to the field of distribution network line parameter identification technology. (See attached document.) Figure 1 The method includes the following steps: Step 1: Obtain voltage and current PMU measurement data at the beginning and end of the distribution network to be identified, and extract the positive sequence voltage phasor and positive sequence current phasor.

[0020] PMU devices are deployed at both the beginning and end of the distribution network line to be identified. The PMU devices collect the three-phase voltage phasors at both ends of the line in real time at a set sampling frequency (e.g., 50Hz). , ), three-phase current phasors ( , Measurement data, and simultaneously obtain current distribution network topology information and line rated resistance through the distribution network dispatch system. Rated reactance .

[0021] The collected PMU measurement data undergoes preprocessing as follows: First, an abnormal jump value is removed from the measurement data using a sliding window method. The length of the sliding window is a preset value N, and the sliding step size is preset to one sampling point. The difference between the measurement data of the current sampling point and the previous sampling point is calculated. If the absolute value of this difference is greater than three times the standard deviation of the data in the previous sliding window, the current sampling point data is determined to be an abnormal jump value, and it is replaced with the mean of the data of the adjacent normal sampling points before and after it to avoid abnormal data affecting the identification accuracy. Second, a Kalman filter algorithm is used to eliminate random noise in the measurement process and improve data quality. Finally, the data collected by different PMU devices are synchronized in time, and the data timestamps are unified to ensure data consistency and validity. The positive sequence components of the three-phase voltage phasors and three-phase current phasors are extracted using the symmetrical component method.

[0022] Step 2: Based on the lumped parameter model of the line, establish the complex equations of the difference between the positive sequence voltage phasors and the positive sequence current phasors at both ends of the line, and expand them into linear observation equations in real form. Vertically stitch together the observation data of multiple consecutive sampling points within the preset sliding window to construct a global observation matrix and a global observation vector.

[0023] Based on the lumped parameter model of the distribution network line, and neglecting the influence of the line-to-ground capacitance (the influence of the line-to-ground capacitance of medium and low voltage distribution network lines is relatively small), the voltage phasor difference and current phasor at both ends of the line satisfy the following relationship: In the formula, The difference between the positive sequence voltage phasors at both ends of the line. , These correspond to the positive sequence voltage phasors at the beginning and end of the line, respectively. The positive sequence current phasor (this application takes the average value of the first and last current phasors). The resistor to be identified in the circuit. For the line reactor to be identified, It is the imaginary unit.

[0024] Transforming the above equations into linear observation equations, and considering the effects of model error, measurement noise, and unknown disturbances, the following parameter identification model is constructed: in, Represents the global observation vector, which consists of the real and imaginary parts of the voltage phasor difference and is used to characterize the core information of the observation data; The parameter vector to be identified is specifically... Including the resistor to be identified in the circuit and reactance ; This is the model error vector, reflecting the error caused by simplification of the line model (such as ignoring capacitance to ground); It measures noise and unknown disturbance vectors, covering random noise and unknown disturbances such as distribution network load fluctuations during PMU measurement.

[0025] Set the sliding window to contain N consecutive sampling points, expand the complex equation into real form, and assume the voltage phasor difference between the two ends of the line. Current phasor ; , Let these represent the real and imaginary parts of the voltage phasor difference, respectively. , Let represent the real and imaginary parts of the line current phasor, respectively, where j is the imaginary unit. The observation vector from a single sample. for Observation matrix A 2x2 matrix N is a positive integer greater than or equal to 2, preferably between 50 and 200. For example, N is 100, but the specific value can be determined based on the PMU sampling frequency and the dynamic characteristics of the line operating conditions. The 2x2 single-observation matrices corresponding to the N sampling points within the sliding window are vertically concatenated to generate a 2Nx2 global observation matrix H. The N 2x1 single-observation vectors are vertically concatenated to generate a 2Nx1 global observation vector Z. The equations in subsequent steps three are calculated using this global observation matrix H and the global observation vector Z, and the total number of observations n in the objective function is 2N.

[0026] Step 3: Construct a parameter identification objective function based on Huber estimation, and use the median absolute deviation algorithm to adaptively update the Huber threshold and weight matrix of the residuals. Iteratively solve the line resistance and reactance until the convergence condition is met.

[0027] The core idea of ​​Huber estimation is to balance the efficiency and robustness of least squares estimation by setting a threshold k to differentiate the residuals. In this application, the objective function for parameter identification based on Huber estimation is: in, The objective function for parameter identification is used to measure the fit between the parameter identification results and the observed data. The goal is to minimize the value of this function through iterative optimization. , represents the residual of the i-th observation, that is, the difference between the observed value and the model prediction value under the current parameters, reflecting the deviation between the observed data and the model; Let be the Huber loss function for the i-th observation, used to differentiate the residuals, balancing the efficiency and robustness of identification. Its expression is: The Huber threshold k is used to distinguish between normal and abnormal residuals; when the absolute value of the residual is... When using squared loss Balancing identification accuracy and computational efficiency; when the absolute value of the residual When using linear loss This helps to suppress the interference of outliers on the identification results.

[0028] Huber threshold The threshold k can be set according to the noise level and outlier distribution of the measurement data, and is usually set to k=1.345 (corresponding to a 95% confidence level). This application further adopts an adaptive method to determine the threshold k, and the specific calculation logic will be explained in detail in subsequent iteration steps.

[0029] The iterative process of parameter identification is shown in the figure below. Figure 2 As shown, the specific steps are as follows: (1) Initialize the parameter vector Take the line's rated parameters ; (2) Calculate the residual sequence under the current parameters. , where t is the number of iterations; (3) Adaptively determine Huber threshold k: Based on the current residual sequence, the median absolute deviation (MAD) is used to robustly estimate the residual standard deviation. The calculation formula is as follows: in: Thus, an adaptive threshold is obtained. .

[0030] The meanings of each indicator are as follows: 1.4826 is the robust estimate standard deviation of the residuals, used to reflect the dispersion of the residuals; 1.4826 is the correction factor, used to convert MAD into a robust estimate equivalent to the ordinary standard deviation. This represents the absolute deviation of the median of the residuals. represents the median operation, which calculates the median of the absolute deviations of the residuals from the median of the residuals, thus avoiding interference from outliers in the standard deviation estimation; c is the preset confidence coefficient (recommended value 1.345), corresponding to a 95% confidence level, used to balance robustness and identification accuracy.

[0031] Simultaneously set upper and lower threshold constraints (e.g.) , To avoid extreme threshold values ​​affecting recognition performance, the Huber loss function is then determined based on the adaptive threshold k, and the weight matrix is ​​calculated. The diagonal elements of the weight matrix are ,in This is the derivative of the Huber loss function, used to dynamically adjust the weights of each observation based on the residual size; when the residual... season Alternatively, a very small positive number can be added to the denominator to prevent computational crashes. The smaller the residual, the greater the weight; the larger the residual, the smaller the weight, further suppressing the influence of outliers.

[0032] (4) Update the parameter vector using the weighted least squares method: The meanings of each indicator are as follows: This is the vector of parameters to be identified after the (t+1)th iteration. Let H be the vector of parameters to be identified in the t-th iteration, and let H be the observation matrix. Let Z be the weight matrix for the t-th iteration (the diagonal elements are the ratio of the derivative of the Huber loss function to the residuals, used to dynamically adjust the weights of each observation), and let Z be the observation vector. This is a matrix inversion operation used to find the optimal parameter solution for weighted least squares.

[0033] (5) Adaptive convergence criterion: Calculate the change in parameters ,in, Let be the parameter change in the t-th iteration. It is the 2-norm, used to measure the degree of difference between the parameter vectors in two iterations; , These are the parameter vectors to be identified in the (t+1)th and tth iterations, respectively; a two-level convergence strategy is adopted, if there are M consecutive iterations (e.g., 3 times)... The price continues to decrease and is below the first convergence threshold. (e.g. 10) -4 If ), then the iteration is terminated early to improve recognition efficiency; if If oscillations continue (e.g., the absolute value of the relative difference between the change in parameters in the current iteration and the change in parameters in the previous iteration exceeds 20% for 5 consecutive iterations), then the first convergence threshold will be tightened to the second convergence threshold. (e.g. 10) -5 Increase the upper limit of the number of iterations appropriately to ensure the identification accuracy; when the convergence condition is met or the maximum number of iterations is reached, stop the iteration and output the current parameter vector as the preliminary identification result; otherwise, return to step (2) to continue the iteration.

[0034] Step 4: Monitor the distribution network topology in real time. When the topology changes, reset the sliding window. After the newly collected measurement data reaches the window length, reconstruct the global observation matrix and global observation vector for identification.

[0035] The system monitors changes in the distribution network topology in real time, and obtains event information such as line switching and distributed generation access / exit through the dispatch system. When a topology change is detected, a parameter correction mechanism is triggered. The specific process is as follows: (1) Update the observation matrix H and observation vector Z of the parameter identification model according to the line connection relationship after the topology change; (2) Clear the original data sliding window, pause parameter identification, and continue until the measurement data collected after the topology change reaches the sliding window length N. Then, use the new data to construct the global observation matrix H and the global observation vector Z to perform parameter identification. (3) Replace the original parameters with the corrected parameters to ensure the accuracy of subsequent analysis and decision-making.

[0036] Step 5: Verify the error and stability of the identified line resistance and reactance, and output the final parameter identification results.

[0037] The identified line resistance R and reactance X are verified. The verification criteria are: the relative error between the parameter identification result and the line rated parameters does not exceed a preset error threshold (e.g., 5%), and the ratio of the range (maximum value minus minimum value) of the parameter identification result over three consecutive sliding window periods to its mean does not exceed a preset stability threshold of 1%. If the verification criteria are met, the final identification result is output and applied to scenarios such as distribution network state estimation, fault diagnosis, and power flow calculation. If the verification criteria are not met, the Huber threshold or iterative parameters are adjusted. A combination of fixed adjustment and adaptive adjustment can be used, where the adaptive adjustment logic is closely linked to the adaptive threshold and convergence judgment logic estimated by Huber in step three. The fixed adjustment method is as follows: if the parameter identification deviation is large and there is obvious outlier interference, the Huber threshold k can be appropriately increased (e.g., from 1.345 to 1.5) to enhance the ability to suppress outliers; if the parameter convergence speed is too slow or the convergence condition is not met, the upper limit of the number of iterations can be increased (e.g., from 50 to 100), and the convergence threshold ε can be fine-tuned (e.g., from 10...). -4 Adjusted to 10 -5 This improves recognition accuracy.

[0038] The adaptive adjustment method is as follows: reuse the robust estimation method of residual standard deviation based on median absolute deviation (MAD) in step three, update the Huber threshold in real time and apply upper and lower limit constraints, and combine the convergence judgment logic in step three to dynamically adjust the iteration step size and convergence threshold according to the parameter convergence rate during the iteration process. After adjustment, re-execute the parameter identification step in step three. In this process, set a maximum number of iterations (e.g., 5 times). If the accuracy requirement is not met after the maximum number of iterations, the adjustment is terminated, an abnormal parameter alarm is output, and the identification result of the previous convergence cycle is taken as the current output.

Claims

1. A method for online identification of distribution network line parameters based on PMU measurement, characterized in that, The method includes the following steps: Acquire voltage and current PMU measurement data at the beginning and end of the distribution network to be identified, and extract positive sequence voltage phasors and positive sequence current phasors; Based on the lumped parameter model of the line, a complex equation for the difference between the positive sequence voltage phasors and the positive sequence current phasors at both ends of the line is established and expanded into a linear observation equation in real form. The observation data of multiple consecutive sampling points within a preset sliding window are vertically spliced ​​together to construct a global observation matrix and a global observation vector. Based on Huber estimation, a parameter identification objective function is constructed. The Huber threshold and weight matrix of the residuals are adaptively updated using the median absolute deviation algorithm. The line resistance and reactance are iteratively solved until the convergence condition is met. The distribution network topology is monitored in real time. When the topology changes, the sliding window is reset. After the newly collected measurement data reaches the window length, the global observation matrix and global observation vector are reconstructed for identification. The identified line resistance and reactance are checked for error and stability, and the final parameter identification results are output.

2. The online identification method for distribution network line parameters based on PMU measurement as described in claim 1, characterized in that, The complex equations for the positive-sequence voltage phasor difference and positive-sequence current phasor at both ends of the line are established and expanded into linear observation equations in real form. The specific formula is as follows: In the formula, Represents the global observation vector; The parameter vector to be identified; This is the model error vector; H represents the measurement noise and unknown disturbance vector; H represents the global observation matrix.

3. The online identification method for distribution network line parameters based on PMU measurement as described in claim 2, characterized in that, The process of obtaining the global observation matrix and global observation vector is as follows: Calculate the positive sequence voltage phasor difference between the two ends of the line. ,in , These are the extracted positive sequence voltage phasors at the beginning and end of the line, respectively. Establish complex equations: ,in For positive sequence current phasors and To identify the line resistance and reactance, The imaginary unit; The positive sequence voltage phasor difference and positive sequence current phasor are expanded into real and imaginary parts. Based on the principle that the real parts are equal to the real parts and the imaginary parts are equal to the imaginary parts, a single observation vector and a single observation matrix for a single sampling are constructed. The global observation vector is generated by vertically concatenating the single observation vectors corresponding to all sampling points within the preset sliding window, and the global observation matrix is ​​generated by vertically concatenating the single observation matrices corresponding to all sampling points.

4. The online identification method for distribution network line parameters based on PMU measurement as described in claim 1, characterized in that, The parameter identification objective function is specifically obtained by summing the Huber loss functions corresponding to the residuals of all observations.

5. The online identification method for distribution network line parameters based on PMU measurement as described in claim 4, characterized in that, The Huber loss function corresponding to the residual is as follows: If the absolute value of the residual is less than or equal to the Huber threshold, then the value of the Huber loss function is half the square of the residual; If the absolute value of the residual is greater than the Huber threshold, then the value of the Huber loss function is the product of the Huber threshold and the absolute value of the residual, minus half the square of the Huber threshold.

6. The online identification method for distribution network line parameters based on PMU measurement as described in claim 1, characterized in that, The process of adaptively updating the Huber threshold and weight matrix of the residuals using the median absolute deviation algorithm is as follows: Based on the residual sequence of the current iteration, the robust estimate of the standard deviation of the residual is obtained by calculating the median absolute deviation; The robust estimation standard deviation is multiplied by a preset confidence coefficient to output the updated Huber threshold, and then truncation constraints are applied to it using preset upper and lower threshold limits. The ratio of the derivative of the Huber loss function to the residual is used as the weight of the corresponding observation to construct the updated weight matrix; when the residual is zero, the corresponding weight is directly output as 1 to prevent division by zero error.

7. The online identification method for distribution network line parameters based on PMU measurement as described in claim 1, characterized in that, The iterative solution for line resistance and reactance specifically involves: The line's rated resistance and rated reactance are used as the initial values ​​for the parameter vector to be identified; Based on the weight matrix, global observation matrix, and global observation vector of the current iteration, the updated parameter vector to be identified is calculated using the weighted least squares formula. Calculate the change in the parameter vector to be identified before and after this iteration update, and perform a two-level convergence determination; The iteration stops when the convergence condition is met or the maximum number of iterations is reached, and the current vector of parameters to be identified is output as the preliminary identification result of the line resistance and reactance.

8. The online identification method for distribution network line parameters based on PMU measurement as described in claim 7, characterized in that, The specific process of the two-stage convergence determination is as follows: The initial convergence threshold is set as the first convergence threshold; if the absolute value of the relative difference between the change in the current iteration parameter and the change in the previous iteration parameter is greater than the preset fluctuation ratio, the convergence threshold is tightened from the first convergence threshold to a smaller second convergence threshold, and the maximum number of iterations is increased simultaneously. Accordingly, the convergence condition specifically includes any one of the following: the change in the parameter vector to be identified decreases continuously for a preset number of consecutive times and is less than the current convergence threshold; the number of iterations reaches the current maximum number of iterations limit.

9. The online identification method for distribution network line parameters based on PMU measurement as described in claim 1, characterized in that, The determination of a change in the topology is based on: obtaining line switching event information or distributed power source access / exit event information in real time through the distribution network dispatching system; when the event information is received, it is determined that the distribution network topology has changed.

10. The online identification method for distribution network line parameters based on PMU measurement as described in claim 1, characterized in that, The specific process of performing error and stability checks on the identified line resistance and reactance, and outputting the final parameter identification result, is as follows: The relative error between the parameter identification results and the line rated parameters is calculated, as well as the ratio of the range of the parameter identification results over multiple consecutive sliding window periods to their mean. If the relative error does not exceed a preset error threshold and the ratio does not exceed a preset stability threshold, the verification is deemed successful, and the current parameter identification result is output as the final result. If the verification fails, then based on the parameter error and convergence status, perform at least one of the following operations: increase the Huber threshold or increase the maximum number of iterations, and re-execute the parameter identification based on Huber estimation. If the number of iterations for parameter identification reaches the preset maximum limit and the verification still fails, the adjustment will be terminated, a parameter anomaly alarm will be output, and the identification result of the previous convergence cycle will be used as the final result.

Images