A power measurement point data threshold anti-interference judgment method and system
By using a Kalman filter model and multi-feature collaborative judgment, combined with cross-validation of historical and adjacent measurement point data, the problem of weak anti-interference capability of power measurement point data is solved, achieving accurate differentiation between interference and real anomalies, and improving the stability and reliability of power system monitoring.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- NARI TECH CO LTD
- Filing Date
- 2026-03-16
- Publication Date
- 2026-06-19
AI Technical Summary
Existing threshold judgment methods for power measurement point data have weak anti-interference capabilities, cannot effectively distinguish between normal fluctuations and external interference, have rigid threshold settings, poor adaptability, and lack noise suppression and verification mechanisms, leading to frequent misjudgments and missed judgments.
Data preprocessing is performed using a Kalman filter model, dynamic thresholds are set, and data types are distinguished through multi-feature collaborative judgment. By combining historical and adjacent measurement point data for cross-verification, accurate distinction between interference and real anomalies can be achieved.
It improves the accuracy and anti-interference capability of power measurement point data threshold judgment, reduces false judgments and missed judgments, ensures the reliability of judgment results, and supports power grid regulation and equipment operation and maintenance.
Smart Images
Figure CN122241471A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the fields of power system automation, data monitoring and anti-interference technology, and in particular to a method and system for judging the threshold of power measurement point data anti-interference based on Kalman filtering. Background Technology
[0002] Power system monitoring data is the core foundation for grid regulation, equipment operation and maintenance, and fault diagnosis. The accuracy of the monitoring data directly determines the safety and stability of power system operation. However, the power system operating environment is complex, and monitoring data is susceptible to various interferences during acquisition and transmission, including electromagnetic interference, high-frequency noise, and transmission link interference. This can lead to abnormal fluctuations and excessive deviations in the monitoring data, affecting its validity.
[0003] Existing methods for determining thresholds in power measurement data often employ fixed or simple statistical thresholds, lacking robust anti-interference capabilities and exhibiting the following shortcomings: 1. Weak anti-interference capability: They cannot effectively distinguish between normal fluctuations, external interference, and genuine anomalies in measurement data, easily misclassifying false fluctuations caused by interference as genuine anomalies, or masking genuine anomalies due to interference, leading to misjudgments or missed judgments; 2. Rigid threshold settings: They fail to dynamically adjust based on the temporal characteristics and fluctuation patterns of the measurement data, resulting in poor adaptability and low universality of thresholds across different types of measurement points; 3. Lack of effective noise suppression processing for the raw data: High-frequency noise and random interference directly affect the accuracy of threshold determination; 4. Lack of effective verification mechanisms for the judgment results: It is difficult to ensure the reliability of the judgment results, potentially misleading power grid control and maintenance work. Summary of the Invention
[0004] Purpose of the invention: The purpose of this invention is to provide a method and system for anti-interference judgment of threshold values for power measurement point data, which solves the problems of power measurement point data being easily interfered with, low threshold judgment accuracy, and difficulty in distinguishing interference from real anomalies. This method can improve the accuracy and anti-interference capability of threshold judgment for power measurement point data, and achieve accurate distinction between interference and real anomalies.
[0005] Technical solution: The power measurement point data threshold anti-interference judgment method of the present invention includes:
[0006] (1) Collect raw data from power system measurement points and perform preprocessing;
[0007] (2) Input the preprocessed measurement point data into the Kalman filter model;
[0008] (3) Based on the type and operation requirements of the power measurement points, set the normal threshold range and fluctuation threshold for the data, and calculate the fluctuation characteristic parameters of the data;
[0009] (4) Establish anti-interference judgment rules, compare the fluctuation characteristic parameters with the preset threshold, and distinguish normal data fluctuation, external interference and real anomaly through multi-feature collaborative judgment;
[0010] (5) If interference or data abnormality is detected, the corresponding early warning and verification mechanism is triggered. If the verification is correct, the early warning information is output; if the verification is abnormal, the filtering parameters or threshold range are adjusted.
[0011] Furthermore, in step (1), the preprocessing includes outlier removal and missing value imputation.
[0012] Furthermore, in step (2), after the measurement point data is input into the Kalman filter model, state prediction, covariance prediction, Kalman gain calculation, state update, and covariance update are performed.
[0013] Furthermore, in step (3), the normal data threshold is determined based on the measurement point design parameters and the statistical characteristics of historical operating data.
[0014] Furthermore, in step (3), the fluctuation threshold includes an instantaneous fluctuation threshold, a continuous fluctuation duration threshold, and a fluctuation frequency threshold.
[0015] Furthermore, in step (4), the fluctuation characteristic parameters include: instantaneous fluctuation amount, continuous fluctuation duration, and fluctuation frequency.
[0016] Furthermore, in step (4), the anti-interference judgment rule is as follows:
[0017] If the filtered data is within the normal threshold range, and the instantaneous fluctuation amount is less than the instantaneous fluctuation threshold, the continuous fluctuation duration is less than the continuous fluctuation duration threshold, and the fluctuation frequency is less than the fluctuation frequency threshold, then it is determined that there is no interference.
[0018] If the filtered data is within the normal threshold range, but the instantaneous fluctuation is greater than the instantaneous fluctuation threshold, the continuous fluctuation duration is less than the continuous fluctuation duration threshold, and the fluctuation frequency is greater than or equal to the fluctuation frequency threshold, it is determined that there is external interference.
[0019] If the filtered data exceeds the normal threshold range, and the instantaneous fluctuation is greater than the instantaneous fluctuation threshold, the continuous fluctuation duration is greater than or equal to the continuous fluctuation duration threshold, and the fluctuation frequency is greater than or equal to the fluctuation frequency threshold, it is determined that there is a real data anomaly.
[0020] If the filtered data exceeds the normal threshold range, but the instantaneous fluctuation amount is less than or equal to the instantaneous fluctuation threshold and the continuous fluctuation duration is less than the continuous fluctuation duration threshold, it is judged as a suspected anomaly.
[0021] Further, in step (5), the verification includes:
[0022] The current filtered data is compared with historical measurement data from the same period, and the deviation value is calculated.
[0023] The filtered data of the current measuring point is compared with the filtered data of adjacent measuring points of the same type, and the deviation value is calculated.
[0024] Furthermore, if both deviation values are ≤ a preset multiple of the average deviation of historical data, it is determined to be a real anomaly or interference; otherwise, it is determined to be a verification anomaly, and the filter parameters or threshold range is returned.
[0025] A power measurement point data threshold anti-interference judgment system, comprising:
[0026] Data acquisition and preprocessing module: used to acquire raw data from power system measurement points and perform outlier removal and missing value filling preprocessing on the raw data;
[0027] Kalman filter modeling and filtering module: used to input preprocessed data into the model for filtering and output filtered data;
[0028] Threshold setting module: Sets the normal threshold range and fluctuation threshold for data according to the type of power measurement point and operating requirements;
[0029] Feature extraction and judgment module: used to calculate the fluctuation feature parameters of the filtered data and to determine whether the data is normal, interfered with, or truly abnormal;
[0030] Early warning and verification module: Used to output the final judgment result and early warning information.
[0031] Beneficial Effects: Compared with existing technologies, this invention has the following significant advantages: 1. It effectively suppresses random interference and high-frequency noise in the measurement point data, restores the true characteristics of the data, and improves the accuracy of threshold judgment from the source; 2. It adopts the quantile method combined with time-series fluctuation characteristics to set dynamic thresholds, solving the problem of poor adaptability of existing fixed thresholds and adapting to the operational needs of different types of power measurement points; 3. It establishes multi-dimensional anti-interference judgment rules, combining characteristic parameters such as instantaneous fluctuations, continuous fluctuation duration, and fluctuation frequency to achieve accurate differentiation between normal data fluctuations, external interference, and real anomalies, reducing misjudgments and omissions; 4. It adds a cross-validation mechanism, comparing historical data with adjacent measurement point data to further improve the reliability of the judgment results, providing accurate and effective data support for power system regulation and equipment operation and maintenance, and significantly improving the stability and practicality of power system measurement point monitoring. Attached Figure Description
[0032] Figure 1 This is a schematic diagram of the system module architecture of the present invention;
[0033] Figure 2 This is a comparison curve of the original data and the filtered data of the voltage measurement points in this invention;
[0034] Figure 3This is a schematic diagram showing the results of the interference judgment of the current measurement point data in this invention. Detailed Implementation
[0035] The technical solution of the present invention will be further described below with reference to the accompanying drawings.
[0036] like Figure 1 As shown, the power measurement point data threshold anti-interference judgment method based on Kalman filtering described in this invention includes the following steps:
[0037] (1) Data acquisition and preprocessing;
[0038] Raw data from power system measurement points are collected, and the raw data is preprocessed to remove obvious outliers that exceed reasonable ranges and fill in missing data to obtain standardized preprocessed data.
[0039] The preprocessing includes outlier removal and missing value imputation: outlier removal uses the 3σ criterion, which identifies data exceeding the range of [μ-3σ, μ+3σ] as obvious outliers and removes them, where μ is the mean of the original data and σ is the standard deviation of the original data; missing value imputation uses linear interpolation, which calculates interpolation based on the valid data adjacent to the missing data to fill in the missing items.
[0040] (2) Input the preprocessed measurement point data into the Kalman filter model, and sequentially execute the steps of state prediction, covariance prediction, Kalman gain calculation, state update, and covariance update to obtain the filtered data, thereby suppressing random interference and high-frequency noise in the measurement point data, such as Figure 2 As shown;
[0041] Based on the time-series correlation and fluctuation patterns of voltage measurement data, the state equation and observation equation are determined using a Kalman filter model. The state equation of the Kalman filter model is as follows:
[0042] ,
[0043] Among them, among them, Let k be the state vector at time k, corresponding to the true value of the measurement point data; The state transition matrix is set as a first-order autoregressive coefficient matrix based on the temporal correlation of the measurement point data and the steady-state operation characteristics of the power system. To control the input matrix, This is the control input at time k-1, which is set to 0 in the power measurement point monitoring scenario. The noise is the process noise, which follows a mean of 0 and a covariance of . Gaussian distribution;
[0044] The observation equation is:
[0045] ;
[0046] in, Let be the observation vector at time k, corresponding to the preprocessed measurement point data; The observation matrix is taken as the identity matrix; To observe the noise, it follows a pattern with a mean of 0 and a covariance of . The Gaussian distribution.
[0047] The execution steps of the Kalman filter model are as follows:
[0048] (21) State prediction: ,in Let be the predicted state vector at time k. This is the updated state vector at time k-1;
[0049] (22) Covariance prediction: ,in Let be the covariance matrix of the predicted state at time k. Let be the covariance matrix of the state after the update at time k-1. This is the transpose of the state transition matrix A;
[0050] (23) Kalman gain calculation: ,in The Kalman gain at time k, The transpose of the observation matrix H, This represents finding the inverse of a matrix.
[0051] (24) Status update: ,in This is the updated state vector at time k, i.e., the filtered data;
[0052] (25) Covariance update: ,in Let be the covariance matrix of the state after the update at time k. identity matrix
[0053] (3) Based on the type and operation requirements of the power measurement points, set the normal threshold range and fluctuation threshold of the data, and calculate the instantaneous fluctuation amount, continuous fluctuation duration and fluctuation frequency of the data in combination with the filtered data;
[0054] The normal data threshold is determined based on the design parameters of the measuring points and the statistical characteristics of historical operating data, and is set using the quantile method as [Q25, Q75], where Q25 is the 25th quantile of historical data and Q75 is the 75th quantile of historical data. The fluctuation threshold includes the instantaneous fluctuation threshold, the continuous fluctuation duration threshold, and the fluctuation frequency threshold, which are set to 1.5 times the average fluctuation of historical data, 3 data acquisition cycles, and 5 times / minute, respectively, achieving dynamic adaptation of the thresholds, which differs from the traditional fixed threshold setting method.
[0055] (4) Establish anti-interference judgment rules, compare the calculated fluctuation characteristic parameters with preset thresholds, and distinguish normal fluctuations, external interferences and real anomalies through multi-feature collaborative judgment to complete the threshold anti-interference judgment, such as Figure 3 As shown; the anti-interference judgment rules are as follows:
[0056] If the filtered data is within the normal threshold range, and the instantaneous fluctuation amount is less than the instantaneous fluctuation threshold, the continuous fluctuation duration is less than the continuous fluctuation duration threshold, and the fluctuation frequency is less than the fluctuation frequency threshold, then the data is judged to be normal and without interference.
[0057] If the filtered data is within the normal threshold range, but the instantaneous fluctuation is greater than the instantaneous fluctuation threshold, the continuous fluctuation duration is less than the continuous fluctuation duration threshold, and the fluctuation frequency is greater than or equal to the fluctuation frequency threshold, then it is determined that there is external interference.
[0058] If the filtered data exceeds the normal threshold range, and the instantaneous fluctuation is greater than the instantaneous fluctuation threshold, the continuous fluctuation duration is greater than or equal to the continuous fluctuation duration threshold, and the fluctuation frequency is greater than or equal to the fluctuation frequency threshold, then it is determined that there is a real data anomaly.
[0059] If the filtered data exceeds the normal threshold range, but the instantaneous fluctuation is less than or equal to the instantaneous fluctuation threshold and the continuous fluctuation duration is less than the continuous fluctuation duration threshold, it is judged as a suspected anomaly and requires further verification.
[0060] The fluctuation characteristic parameters include: instantaneous fluctuation amount, which is the absolute value of the difference between the current filtered data and the filtered data at the previous moment; continuous fluctuation duration, which is the number of consecutive acquisition cycles in which the data fluctuation amount exceeds the instantaneous fluctuation threshold; and fluctuation frequency, which is the number of times the data fluctuation amount exceeds the instantaneous fluctuation threshold per unit time. The three-dimensional features comprehensively characterize the data fluctuation characteristics, providing support for distinguishing between interference and real anomalies.
[0061] (5) If interference or data abnormality is detected, the corresponding warning and verification mechanism is triggered. The reliability of the judgment result is ensured through double cross-verification. If the verification is correct, the warning information is output. If the verification is abnormal, the filter parameters or threshold range are readjusted.
[0062] The cross-validation method combines historical data comparison with adjacent measurement point data comparison: the current filtered data is compared with historical measurement point data from the same period, and the deviation value is calculated; the current filtered measurement point data is compared with filtered data from adjacent measurement points of the same type, and the deviation value is calculated; if both deviation values are ≤ 1.2 times the average deviation of historical data, the verification is correct and it is determined to be a real anomaly or interference; otherwise, it is determined to be a verification anomaly, and the process returns to readjust the Q and R parameters or threshold range of the Kalman filter model, and the filtering and judgment steps are re-executed to form a closed-loop optimization mechanism, further improving the reliability of the judgment results.
[0063] A threshold anti-interference judgment system for power measurement point data based on Kalman filtering, comprising:
[0064] Data Acquisition and Preprocessing Module: The module uses a data acquisition device to collect raw data from power system measurement points, and performs outlier removal and missing value filling operations through the preprocessing unit to output standardized preprocessed data.
[0065] Kalman filter modeling and filtering module: It has a built-in Kalman filter modeling unit and a filtering unit. The modeling unit can adaptively adjust the model parameters according to the measurement point type, and the filtering unit performs state prediction, covariance prediction and other steps to output filtered data.
[0066] Threshold setting module: Supports two modes: manual setting and automatic setting. In automatic setting mode, the threshold range can be automatically calculated and set based on the statistical characteristics of historical data. In manual setting mode, the threshold can be adjusted by the user according to the design parameters of the measurement point.
[0067] Feature extraction and judgment module: Extracts the fluctuation feature parameters of the filtered data, combines them with preset anti-interference judgment rules, completes the data state judgment, and outputs the judgment result;
[0068] Early warning and verification module: It has a built-in early warning unit and a verification unit. The early warning unit outputs corresponding early warning information based on the judgment result, and the verification unit uses historical data comparison and adjacent measurement point comparison to perform cross-verification to ensure the reliability of the judgment result.
[0069] This system can be directly integrated into the power system monitoring platform, adapting to various types of measurement point data such as voltage, current, and power, enabling real-time anti-interference judgment and early warning of measurement point data, and improving the intelligence level of power system monitoring.
[0070] Example 1
[0071] (1) Data acquisition and preprocessing
[0072] Raw data from voltage measurement points in the power system were collected at a frequency of 1 time / second for 1 hour, resulting in 3600 sets of raw data. The raw data underwent preprocessing: The 3σ criterion was used to remove obvious outliers. The mean μ = 10kV and standard deviation σ = 0.2kV were calculated. Data exceeding the range [9.4kV, 10.6kV] were identified as obvious outliers, and 12 sets of outliers were removed. Linear interpolation was used to fill in missing data. Assuming 8 sets of missing data existed during the collection process, linear interpolation was performed using the two adjacent valid sets of data to fill in the missing data, resulting in 3600 sets of standardized preprocessed data.
[0073] (2) Kalman filtering
[0074] The preprocessed 3600 sets of voltage measurement data were sequentially input into the Kalman filter model, and the following steps were performed:
[0075] (21) State prediction: Based on the updated state vector from the previous time step, predict the state vector at the current time step.
[0076] (22) Covariance prediction: Predict the covariance matrix of the current state vector;
[0077] (23) Kalman gain calculation: Calculate the Kalman gain at the current time, which is used to balance the weights of the predicted and observed values;
[0078] (24) Status update: The state vector at the current moment is updated by combining the observed values (preprocessed data) to obtain the filtered data;
[0079] (25) Covariance update: Update the covariance matrix of the current state vector to prepare for filtering in the next time step.
[0080] Through the above steps, 3600 sets of filtered voltage measurement data are obtained, which can effectively suppress high-frequency noise and random interference in the original data, making the data fluctuations more stable.
[0081] (3) Threshold setting
[0082] Based on the design parameters of the voltage measuring point (rated voltage 10kV) and historical operating data (voltage measuring point data for the past month), the threshold range is set as follows:
[0083] Normal data threshold: Using the quantile method, the 25th quantile of historical data is calculated as Q25 = 9.8kV and the 75th quantile as Q75 = 10.2kV. Therefore, the normal threshold range is set as [9.8kV, 10.2kV].
[0084] Fluctuation threshold: The average fluctuation of historical data is calculated to be 0.05kV, so the instantaneous fluctuation threshold is set to 0.075kV (1.5×0.05kV); the continuous fluctuation duration threshold is set to 3 acquisition cycles (3 seconds); the fluctuation frequency threshold is set to 5 times / minute.
[0085] (4) Anti-interference judgment
[0086] Establish anti-interference judgment rules, compare fluctuation characteristic parameters with preset thresholds, and distinguish normal data fluctuations, external interference and real anomalies through multi-feature collaborative judgment.
[0087] Based on the filtered data, calculate the fluctuation characteristic parameters for each data set:
[0088] Instantaneous fluctuation: , which is the absolute value of the difference between the filtered data at the current time and the previous time.
[0089] Continuous fluctuation duration: The number of consecutive acquisition cycles with instantaneous fluctuations > 0.075kV;
[0090] Fluctuation frequency: The number of times the instantaneous fluctuation is greater than 0.075kV per minute.
[0091] Based on the anti-interference judgment rules, each set of data is judged as follows:
[0092] At a certain moment, the filtered data is 10.0kV (within the normal threshold range), the instantaneous fluctuation is 0.04kV≤0.075kV, the continuous fluctuation duration is 1 second<3 seconds, and the fluctuation frequency is 3 times / minute<5 times / minute. The data is judged to be normal and without interference.
[0093] At a certain moment, the filtered data is 10.1kV (within the normal threshold range), the instantaneous fluctuation is 0.08kV > 0.075kV, the continuous fluctuation duration is 2 seconds < 3 seconds, and the fluctuation frequency is 6 times / minute ≥ 5 times / minute, indicating that there is external interference.
[0094] At a certain moment, the filtered data is 10.3kV (exceeding the normal threshold range), the instantaneous fluctuation is 0.09kV > 0.075kV, the continuous fluctuation duration is 4 seconds ≥ 3 seconds, and the fluctuation frequency is 7 times / minute ≥ 5 times / minute. It is determined that there is a real anomaly in the data.
[0095] At a certain moment, the filtered data was 10.3kV (exceeding the normal threshold range), the instantaneous fluctuation was 0.06kV ≤ 0.075kV, and the continuous fluctuation duration was 2 seconds < 3 seconds. It was judged as a suspected anomaly and further verification was required.
[0096] (5) Early warning and cross-validation
[0097] For data identified as having external interference, genuine anomalies, or suspected anomalies, an early warning mechanism is triggered, and corresponding early warning information (interference warning, anomaly warning, suspected anomaly warning) is output; cross-validation is performed on suspected and genuine anomaly data.
[0098] (51) Historical data comparison: Compare the current filtered data (10.3kV) with the historical data of the same period (voltage data of 10.05kV at the same time last week), and calculate the deviation value = 0.25kV; the average deviation of historical data is 0.2kV, 0.25kV≤1.2×0.2kV=0.24kV, the comparison is qualified;
[0099] (52) Comparison of adjacent measuring points: Compare the filtered data (10.3kV) of the current voltage measuring point with the filtered data (10.28kV) of the adjacent voltage measuring points of the same type. Calculate the deviation value = 0.02kV ≤ 0.24kV, and the comparison is qualified;
[0100] (53) Verification conclusion: If both deviation values meet the requirements, the verification is correct, and the data is determined to be a real anomaly. An anomaly warning message is output. If either deviation value exceeds the requirements, it is determined to be a verification anomaly. The Kalman filter model is adjusted to Q=0.008 and R=0.04, and the filtering and judgment steps are re-executed.
[0101] Example 2
[0102] Taking current measurement data from power system points as an example:
[0103] The raw data from the current measurement points was collected at a frequency of 1 time / second for 1 hour, totaling 3600 data sets. The raw data included high-frequency fluctuations caused by electromagnetic interference and some missing data. The method of this invention was used for interference assessment.
[0104] (1) Preprocessing: Outliers were removed using the 3σ criterion, μ=50A, σ=1A, and the range of outliers was [47A, 53A]. 15 groups of outliers were removed. 6 groups of missing data were filled using linear interpolation to obtain standardized preprocessed data.
[0105] (2) Kalman filter model parameters: state transition matrix A=0.97, observation matrix H=1, process noise covariance Q=0.012, observation noise covariance R=0.055, initial value of state vector The mean of the first 5 preprocessed data sets is 50.02A, and the initial value of the state covariance is... .
[0106] (3) Filtering: Input the preprocessed data into the model and perform the filtering step to obtain the filtered data. The fluctuation amplitude of the filtered data is significantly reduced and high-frequency interference is effectively suppressed.
[0107] (4) Threshold setting: normal threshold range [48.5A, 51.5A] (Q25=48.5A, Q75=51.5A), instantaneous fluctuation threshold = 0.15A (historical average fluctuation amount 0.1A×1.5), continuous fluctuation duration threshold = 3 seconds, fluctuation frequency threshold = 5 times / minute.
[0108] (5) Anti-interference judgment: The filtered data for a certain period is 51.0A (within the normal threshold), the instantaneous fluctuation is 0.16A > 0.15A, the continuous fluctuation duration is 2 seconds < 3 seconds, and the fluctuation frequency is 6 times / minute ≥ 5 times / minute. It is determined that there is external interference and an interference warning is output. The filtered data for another period is 52.0A (exceeding the normal threshold), the instantaneous fluctuation is 0.17A > 0.15A, the continuous fluctuation duration is 4 seconds ≥ 3 seconds, and the fluctuation frequency is 7 times / minute ≥ 5 times / minute. The cross-validation is qualified and it is determined to be a real anomaly. An anomaly warning is output.
[0109] Compared with existing fixed threshold judgment methods, the existing methods have a false positive rate of 12.3% and a false negative rate of 8.7%; the present method has a false positive rate of 2.1% and a false negative rate of 1.5%, thus improving the judgment accuracy and anti-interference ability.
Claims
1. A method for judging the anti-interference threshold of power measurement point data, characterized in that, include: (1) Collect raw data from power system measurement points and perform preprocessing; (2) Input the preprocessed measurement point data into the Kalman filter model; (3) Based on the type and operation requirements of the power measurement points, set the normal threshold range and fluctuation threshold for the data, and calculate the fluctuation characteristic parameters of the data; (4) Establish anti-interference judgment rules, compare the fluctuation characteristic parameters with the preset threshold, and distinguish normal data fluctuation, external interference and real anomaly through multi-feature collaborative judgment; (5) If interference or data abnormality is detected, the corresponding early warning and verification mechanism is triggered. If the verification is correct, an early warning message is output. If the verification fails, adjust the filtering parameters or threshold range.
2. The method for judging the anti-interference threshold of power measurement point data according to claim 1, characterized in that, In step (1), the preprocessing includes outlier removal and missing value imputation.
3. The method for judging the anti-interference threshold of power measurement point data according to claim 1, characterized in that, In step (2), after the measurement point data is input into the Kalman filter model, state prediction, covariance prediction, Kalman gain calculation, state update and covariance update are performed.
4. The method for judging the anti-interference threshold of power measurement point data according to claim 1, characterized in that, In step (3), the normal data threshold is determined based on the measurement point design parameters and the statistical characteristics of historical operating data.
5. The method for judging the anti-interference threshold of power measurement point data according to claim 1, characterized in that, In step (3), the fluctuation threshold includes the instantaneous fluctuation threshold, the continuous fluctuation duration threshold, and the fluctuation frequency threshold.
6. The method for judging the anti-interference threshold of power measurement point data according to claim 1, characterized in that, In step (4), the fluctuation characteristic parameters include: instantaneous fluctuation amount, continuous fluctuation duration, and fluctuation frequency.
7. The method for judging the anti-interference threshold of power measurement point data according to claim 1, characterized in that, In step (4), the anti-interference judgment rule is as follows: If the filtered data is within the normal threshold range, and the instantaneous fluctuation amount is less than the instantaneous fluctuation threshold, the continuous fluctuation duration is less than the continuous fluctuation duration threshold, and the fluctuation frequency is less than the fluctuation frequency threshold, then it is determined that there is no interference. If the filtered data is within the normal threshold range, but the instantaneous fluctuation is greater than the instantaneous fluctuation threshold, the continuous fluctuation duration is less than the continuous fluctuation duration threshold, and the fluctuation frequency is greater than or equal to the fluctuation frequency threshold, it is determined that there is external interference. If the filtered data exceeds the normal threshold range, and the instantaneous fluctuation is greater than the instantaneous fluctuation threshold, the continuous fluctuation duration is greater than or equal to the continuous fluctuation duration threshold, and the fluctuation frequency is greater than or equal to the fluctuation frequency threshold, it is determined that there is a real data anomaly. If the filtered data exceeds the normal threshold range, but the instantaneous fluctuation amount is less than or equal to the instantaneous fluctuation threshold and the continuous fluctuation duration is less than the continuous fluctuation duration threshold, it is judged as a suspected anomaly.
8. The method for judging the anti-interference threshold of power measurement point data according to claim 1, characterized in that, In step (5), the verification includes: The current filtered data is compared with historical measurement data from the same period, and the deviation value is calculated. The filtered data of the current measuring point is compared with the filtered data of adjacent measuring points of the same type, and the deviation value is calculated.
9. The method for judging the anti-interference threshold of power measurement point data according to claim 8, characterized in that, If both deviation values are less than or equal to a preset multiple of the average deviation of historical data, the data is determined to be a genuine anomaly or interference; otherwise, it is determined to be a verification anomaly, and the filter parameters or threshold range is returned.
10. A power measurement point data threshold anti-interference judgment system, characterized in that, include: Data acquisition and preprocessing module: used to acquire raw data from power system measurement points and perform outlier removal and missing value filling preprocessing on the raw data; Kalman filter modeling and filtering module: used to input preprocessed data into the model for filtering and output filtered data; Threshold setting module: Sets the normal data threshold range and fluctuation threshold according to the type of power measurement point and operating requirements; Feature extraction and judgment module: used to calculate the fluctuation feature parameters of the filtered data and to determine whether the data is normal, interfered with, or truly abnormal; Early warning and verification module: used to output the final judgment result and early warning information.