A compensation algorithm for live detection results of distribution gap surge arresters

Abstract

This invention relates to the field of live-line detection technology, specifically a compensation algorithm for live-line detection results of distribution gap surge arresters. The algorithm includes the following steps: acquiring voltage signals and dividing them into periodic segments; extracting waveform features to generate a marker set; identifying trend reversals to construct a list of abrupt change segments; analyzing the distance between intersection points to form an anomaly set; tracking trend offsets to determine correction data; mapping a reference path to reconstruct the reference baseline; and completing the conduction state correction record. In this invention, by processing the detection data of the operating state of the distribution gap surge arrester, continuous periodic waveform segment markers are constructed. Combining the reversal trend path and slope abrupt change features, potential interference is identified and abnormal fluctuation regions are formed. Through a pre-window and trend line segment construction, the voltage recovery offset trend is analyzed, the start and end points of the offset are located, and the reference reference path is corrected. This effectively improves the reliability of the detection data, the accuracy of state judgment, and the adaptability to operating conditions.

Description

Technical Field

[0001] This invention relates to the field of live-line testing technology, and in particular to a compensation algorithm for the live-line testing results of a distribution gap surge arrester. Background Technology

[0002] Live-line detection technology refers to the technical system for condition sensing and performance evaluation of electrical equipment under the operating conditions of power systems. Its core aspects include monitoring electrical parameters, identifying electrical characteristics, assessing insulation status, and judging fault symptoms for key equipment such as transmission lines, distribution devices, transformers, and surge arresters without affecting normal operation. This technical field aims to improve the operational safety and maintenance efficiency of power equipment, encompassing sensing technology, online measurement technology, data compensation analysis, and operating condition identification methods. Through real-time detection of parameters such as equipment voltage, current, leakage current, and electric field strength, it achieves operational status monitoring and anomaly early warning for electrical equipment. Specifically, the compensation algorithm for traditional live-line detection results of distribution gap surge arresters refers to the method of correcting the leakage current signal collected by the gap surge arrester under operating conditions in the distribution system. This addresses the technical issue of detection data errors caused by external influencing factors such as voltage fluctuations, temperature changes, and environmental humidity. The traditional method for compensating for live-line testing results of distribution gap surge arresters is a static amplitude correction method. This method normalizes the leakage current under different environmental parameters and constructs a static empirical model based on the relationship curve between the effective value of the leakage current at power frequency and the gap voltage to correct for environmental interference.

[0003] Existing technologies rely on static amplitude correction methods to normalize leakage current signals when dealing with detection errors caused by external interference factors. This ignores dynamic characteristics such as periodic fluctuations, trend reversals, and voltage drift during operation. Under conditions of frequent voltage disturbances or sudden environmental changes, these technologies cannot capture instantaneous offsets and changing trends, resulting in distorted benchmark judgments and incorrect conduction identification. For example, when load changes cause waveform fluctuations, static models cannot effectively adapt to the changing path during the voltage recovery phase, leading to the risk of misjudgment or omission of detection results. Summary of the Invention

[0004] To address the technical problems existing in the prior art, this invention provides a compensation algorithm for the live detection results of a distribution gap surge arrester, comprising the following steps:

[0005] To achieve the above objectives, the present invention adopts the following technical solution: a compensation algorithm for the live detection results of a distribution gap surge arrester, comprising the following steps:

[0006] S1: Obtain the power frequency voltage signal of the distribution gap surge arrester during online operation, perform time synchronization processing in the voltage recording sequence with a fixed sampling interval, take the zero-crossing point of the whole cycle voltage as the cycle segmentation point, classify it into rising segment and falling segment according to the waveform change direction, and generate a periodic waveform segmentation mark set;

[0007] S2: Based on the periodic waveform segmentation marker set, extract the reversal nodes of multiple segments with changing directions, construct a slope change sequence between nodes, determine whether the number of continuous reversal trends in the connection path exceeds the slope change identification limit, and generate a list of interference candidate change segments.

[0008] S3: Based on the time nodes in the list of candidate mutation segments for interference, configure a pre-observation window of fixed length before the candidate segment, extract the voltage zero-crossing intersection points within the window and record the interval time, sort the interval time in chronological order, and generate a set of abnormal interference stability segments.

[0009] S4: Call the set of abnormal interference stability sections, construct an extended voltage recovery monitoring window at the end of the periodic section, extract the voltage change trend points of the recovery section, compare the trend direction of the real-time periodic recovery section with that of the previous periodic recovery section, and obtain the baseline offset correction section data.

[0010] As a further aspect of the present invention, the periodic waveform segmentation marker set includes waveform amplitude category, segmentation boundary point position, and change trend type; the interference candidate mutation segment list includes mutation start and end time, trend reversal path, and mutation amplitude characteristics; the interference stability abnormal segment set includes intersection spacing variation characteristics, peak difference distribution, and time variation trend; and the baseline offset correction segment data includes offset start and end time sequence, trend line fitting parameters, and baseline deviation amplitude.

[0011] As a further aspect of the present invention, the specific steps of S1 are as follows:

[0012] S101: Acquire the power frequency voltage signal of the distribution gap surge arrester during online operation, perform time synchronization processing on the signal sequence according to a fixed sampling interval, identify the complete power frequency voltage cycle, and perform point-by-point detection of the amplitude in the voltage sequence in combination with the sign change of the voltage change direction. Locate the zero-crossing point in the alternating positive and negative change interval, use the zero-crossing point as the segmentation reference point, divide the voltage signal segment with complete periodic structure, and generate a period segmentation sequence set.

[0013] S102: Based on the period segmentation sequence set, call the voltage waveform data within the period segment, extract all amplitude sample points within the period using an equal-interval indexing method, and make a directional judgment on the amplitude difference between adjacent sample points within the same period. Sample points with the same continuous amplitude growth direction are assigned to the same change segment, and are classified into rising segment or falling segment respectively based on the positive and negative directions of the amplitude difference, generating a waveform directional segmentation set.

[0014] S103: Call the rising and falling segment data in the waveform directionality division set, assign interval label numbers to the corresponding segments based on the start and end ranges determined by the positions of the first and last sample points within the segment, and aggregate the classified waveform segments within the period into corresponding sequence label units to generate a periodic waveform segment label set.

[0015] As a further aspect of the present invention, the specific steps of S2 are as follows:

[0016] S201: Based on the periodic waveform segmentation marker set, sequentially retrieve the marked rising and falling segment sequences in the periodic segment, extract the connection nodes where the change direction is reversed between adjacent waveform segments, and record the position index of the node in the original periodic segment to generate a waveform reversal node sequence set.

[0017] S202: Call the index position between adjacent inverted nodes in the waveform inversion node sequence set, calculate the amplitude slope of the interval by dividing the amplitude change by the corresponding sample point interval within the node interval, arrange the nodes according to the order of nodes, mark the number of inversions of the positive and negative slope between adjacent nodes, and generate a slope change trend sequence group.

[0018] S203: Based on the number of node sequences of continuous reversal paths in the slope change trend sequence group, count the total number of nodes with continuous reversal in direction in each reversal path, and compare it with the set slope change identification limit. When the number of nodes exceeds the slope change identification limit, mark the waveform interval within the coverage area of ​​the corresponding path, aggregate the segment index range that meets the conditions, and generate a list of interference candidate change segments.

[0019] As a further aspect of the present invention, the specific steps of S3 are as follows:

[0020] S301: Based on the time node position in the list of interference candidate mutation segments, configure a fixed-length pre-observation window, and call the original voltage waveform data of the corresponding time period within the observation window, retrieve the amplitude sign information of continuous positive and negative alternation in the waveform, determine the zero-crossing intersection position, and record the number of sampling points between adjacent intersections in sequence. After converting to time units, sort them according to the order of appearance to generate a zero-crossing intersection time interval sequence.

[0021] S302: Call the voltage waveform data within the window corresponding to the zero-crossing intersection time interval sequence, identify the peak and trough points of amplitude in the zero-crossing period segment according to the extreme value detection method, extract the peak amplitude of the period and sort it in time order to generate a peak amplitude difference sequence set;

[0022] S303: Using the peak amplitude difference sequence set, call the corresponding zero-crossing intersection time interval sequence of two adjacent observation windows, compare the difference of continuous values ​​in the intersection time interval sequence between the two adjacent observation windows, compare whether the change direction of the corresponding position in the peak difference sequence is consistent, and when the continuous change trend is consistent and exceeds the set number of judgments, record the corresponding time period of the window as an abnormal segment and generate a set of abnormal interference stability segments.

[0023] As a further aspect of the present invention, the number of determinations is set to be no less than three.

[0024] The difference comparison between the corresponding zero-crossing intersection time interval sequences in two adjacent observation windows is limited to the absolute value of the time difference between adjacent intervals being no less than 1.5 milliseconds;

[0025] The method for determining the consistency of the change direction is to determine that the change direction of the same position in three or more consecutive peak difference sequences is the same, and the change direction of the difference value at the corresponding position in the corresponding intersection time interval sequence is consistent.

[0026] As a further aspect of the present invention, the specific steps of S4 are as follows:

[0027] S401: Call the end time point of the cycle of the abnormal segment in the set of abnormal interference stability segments, extend it by a fixed length to construct a voltage recovery monitoring window, extract continuous voltage amplitude sample points in the voltage recovery monitoring window, construct the recovery segment voltage data sequence according to the time order, identify the position of sign change, and generate a recovery segment trend point series dataset.

[0028] S402: Based on the recovery segment trend point series dataset, call the real-time periodic recovery segment and the recovery segment data sequence of the corresponding position of the previous period respectively, and compare the direction node columns of the two in order. When the continuous direction change shows an offset trend, locate the position where the trend first changes direction as the starting point, detect the end position of the continuous direction deviation as the ending point, connect the starting point and the ending point to construct a straight line segment, and generate a trend offset line segment sequence.

[0029] S403: Call the trend offset line segment sequence and the original reference baseline in the real-time period to perform point-to-point coincidence judgment, calculate the vertical deviation value of the reference baseline within the coverage of the trend offset line segment, extract the range of the deviation value that continuously exceeds the set deviation threshold, mark the waveform segment corresponding to the deviation over-limit range as the correction segment, and obtain the baseline offset correction segment data.

[0030] As a further aspect of the present invention, the set deviation threshold is not less than 5% of the average amplitude in the recovery segment voltage data sequence;

[0031] The time interval between the starting point and the ending point of any trend offset line segment in the trend offset line segment sequence shall not be less than two complete voltage cycles.

[0032] The baseline offset correction segment data includes only waveform segments whose deviation values ​​continuously exceed the set deviation threshold for more than ten sampling points.

[0033] As a further aspect of the present invention, the algorithm further includes step S5:

[0034] S5: Based on the baseline offset correction section data, the live determination baseline in the real-time detection cycle is mapped, the starting point of the mapping path is extracted as the reference value of the previous cycle, and the ending point is the offset point in the middle of the recovery segment trend line. Segmented displacement fitting is performed in the path to reconstruct the reference baseline of the detection cycle. The correction baseline is called in the original detection record for determining whether the surge arrester is conducting to remark the conduction state and generate a compensated live state correction record.

[0035] The compensated live state correction record includes the correction baseline trajectory, conduction state change flag, and correction judgment timestamp.

[0036] As a further aspect of the present invention, the specific steps of S5 are as follows:

[0037] S501: Based on the time interval corresponding to the baseline offset correction section data, call the original charged judgment baseline data within the real-time detection cycle, determine the starting point of the mapping processing within the correction area as the reference amplitude at the same position in the previous cycle, and locate the endpoint at the offset point amplitude corresponding to the middle position of the real-time recovery segment trend line. Perform linear interpolation between the starting and ending points to construct a transition path and generate a mapping path point series dataset.

[0038] S502: Call the interpolation nodes between the start and end points in the mapped path point series dataset, divide it into several continuous sub-segments according to a fixed interval, perform linear fitting processing within the sub-segments, align the nodes within the sub-segments with the original reference baseline, determine the fitting displacement value of each segment through the residual trough value, and generate a reconstructed reference baseline data sequence.

[0039] S503: Synchronously match the reconstructed reference baseline data sequence with the original conduction state detection record, compare the amplitude of the sampling point with the position of the reconstructed baseline in the detection record, re-mark the comparison result as whether it is in the conduction state or non-conduction state, rewrite the conduction mark sequence according to the judgment state of the sampling point, and generate the compensated charged state correction record.

[0040] Compared with the prior art, the advantages and positive effects of the present invention are as follows:

[0041] In this invention, by processing the detection data of the operating status of the distribution gap surge arrester, a continuous periodic waveform segmentation mark is constructed. Combining the reversal trend path and slope change characteristics, potential interference is identified and abnormal fluctuation areas are formed. By constructing the voltage recovery offset trend through the pre-window and trend line segment, the offset start and end points are located and the reference reference path is corrected. Real-time correction of the energized state and compensation judgment of the conduction state are completed. This enables the stability analysis, trend change tracking, baseline drift correction and accurate calibration of detection data under dynamic fluctuations and environmental interference, effectively improving the reliability of detection data, the accuracy of state judgment and the adaptability of operating conditions. Attached Figure Description

[0042] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0043] Figure 1 This is a schematic diagram of the steps of the present invention;

[0044] Figure 2 This is a detailed schematic diagram of S1 of the present invention;

[0045] Figure 3 This is a detailed schematic diagram of S2 of the present invention;

[0046] Figure 4 This is a detailed schematic diagram of S3 of the present invention;

[0047] Figure 5 This is a detailed schematic diagram of S4 of the present invention;

[0048] Figure 6 This is a detailed schematic diagram of S5 of the present invention. Detailed Implementation

[0049] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0050] In embodiments of the present invention, words such as "exemplarily," "for example," etc., are used to indicate that something is an example, illustration, or description. Any embodiment or design described as "exemplary" in the present invention should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of the word "exemplary" is intended to present the concept in a concrete manner. Furthermore, in embodiments of the present invention, the meaning expressed by "and / or" can be both, or either one.

[0051] In the embodiments of this invention, the terms "image" and "picture" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning. Similarly, the terms "of," "corresponding (relevant)," and "corresponding" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning.

[0052] In this embodiment of the invention, sometimes a subscript such as W1 may be written in a non-subscript form such as W1. When the difference is not emphasized, the meaning they express is the same.

[0053] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0054] Please see Figure 1 This invention provides a compensation algorithm for the live detection results of a distribution gap surge arrester, comprising the following steps:

[0055] S1: Obtain the power frequency voltage signal of the distribution gap surge arrester during online operation, perform time synchronization processing in the voltage recording sequence with a fixed sampling interval, take the zero-crossing point of the whole cycle voltage as the cycle segmentation point, divide the continuous cycle segment, extract the waveform amplitude fluctuation nodes in the cycle segment, classify them into rising segment and falling segment according to the waveform change direction, and generate a cycle waveform segmentation mark set.

[0056] S2: Based on the periodic waveform segmentation marker set, extract the reversal nodes of the change direction of multiple segments, construct the slope change sequence between the nodes, perform connection operation on the trend path of continuous reversal of slope change direction, determine whether the number of continuous reversal trends in the connection path exceeds the slope change abruptness identification limit, mark the area as potential interference candidate segment, and generate a list of interference candidate abruptness segments.

[0057] S3: Based on the time nodes in the list of candidate mutation segments for interference, configure a fixed-length pre-observation window before the candidate segment, extract the voltage zero-crossing intersection points in the window and record the interval time, sort the interval time in chronological order, and sort and combine the amplitude peak points in the window. Compare the intersection point spacing and peak point difference sequence in two consecutive observation windows, identify the changing trend of the spacing variation sequence, and generate a set of abnormal interference stability segments.

[0058] S4: Call the set of abnormal interference stability sections, build an extended voltage recovery monitoring window at the end of the cycle section, extract the voltage change trend points of the recovery section, compare the trend direction of the real-time cycle recovery section with the previous cycle recovery section, capture the time position of the trend offset start point and end point, build an offset trend line segment between the two positions, and obtain the baseline offset correction section data based on the overlap deviation range between the trend line segment and the original reference baseline in the real-time cycle.

[0059] S5: Based on the baseline offset correction section data, the live determination baseline in the real-time detection cycle is mapped, the starting point of the mapping path is extracted as the reference value of the previous cycle, and the ending point is the offset point in the middle of the recovery segment trend line. Segmented displacement fitting is performed in the path to reconstruct the reference baseline of the detection cycle. The correction baseline is called in the original detection record for determining whether the surge arrester is conducting to remark the conduction state and generate the compensated live state correction record.

[0060] The periodic waveform segmentation marker set includes waveform amplitude category, segment boundary point location, and change trend type; the interference candidate mutation segment list includes mutation start and end time, trend reversal path, and mutation amplitude characteristics; the interference stability anomaly segment set includes intersection spacing variation characteristics, peak difference distribution, and time variation trend; the baseline offset correction segment data includes offset start and end time sequence, trend line fitting parameters, and baseline deviation amplitude; and the compensated charged state correction record includes the corrected baseline trajectory, conduction state change flag, and correction judgment timestamp.

[0061] Please see Figure 2 The specific steps of S1 are as follows:

[0062] S101: Acquire the power frequency voltage signal of the distribution gap surge arrester during online operation, perform time synchronization processing on the signal sequence according to a fixed sampling interval, identify the complete power frequency voltage cycle, and perform point-by-point detection of the amplitude in the voltage sequence in combination with the sign change of the voltage change direction. Locate the zero-crossing point in the alternating positive and negative change interval, use the zero-crossing point as the segmentation reference point, divide the voltage signal segment with complete periodic structure, and generate a period segmentation sequence set.

[0063] It is necessary to collect the actual operating power frequency voltage waveform data through voltage sensors (such as capacitive voltage transformers CVT or resistive voltage dividers) installed at the surge arrester connection nodes. The sampling frequency must meet the Nyquist criterion, that is, at least twice the grid frequency (50Hz or 60Hz). It is recommended to set the sampling frequency to above 1kHz to obtain complete signal details. The acquired voltage signal data stream is synchronized in time according to a fixed sampling interval (such as Δt=1ms), that is, the timestamp is used as an index to ensure that the voltage value sequence has equal time interval accuracy, ensuring the accuracy of subsequent period division. After acquiring a complete signal, the period structure of the voltage signal is determined by a period identification algorithm (such as based on FFT or peak interval method), and a complete power frequency period (20) is selected. One ms is a period (if the sampling rate is 1kHz, then there are 20 sample points) as the period determination template; the original signal sequence is traversed, and a sliding window mechanism is used for period matching. When the sign of the voltage change direction between consecutive sampling points changes, that is, from positive to negative or from negative to positive, it indicates that the signal has crossed zero. The zero-crossing point is accurately located by interpolation. For example, if the voltage of two points is +5V and -3V by linear interpolation, the zero-crossing point can be calculated to be about 5 / (5+3)=0.625 sampling intervals away from the +5V point. That is, the zero-crossing point is located at 62.5% between the current sampling point and the next sampling point. This point is used as the period segmentation point to divide the signal into segments with complete periodic structures. The entire signal stream is periodically segmented to generate a period segmentation sequence set.

[0064] S102: Based on the period segmentation sequence set, the voltage waveform data within the period segment is called, and all amplitude sample points within the period are extracted using an equal-interval indexing method. Within the same period, the amplitude difference between adjacent sample points is determined directionally. Sample points with the same continuous amplitude growth direction are assigned to the same change segment, and the positive and negative directions of the amplitude difference are combined to classify them into rising segments or falling segments, generating a waveform directional segmentation set.

[0065] Read the voltage waveform data for each period. Assuming a period of 20ms and a sampling rate of 1kHz, one period contains 20 data points. Use an equally spaced indexing method (e.g., take a value every 2 sampling points) to extract all amplitude sample points, forming a periodic sample sequence Vi={v1, v2, ..., vn}. Then, compare the amplitude difference Δv=v between adjacent sample points (vi, vi+1). i+1 -v iThe system determines the positive or negative direction. If Δv > 0, it indicates an upward trend in the current segment; if Δv < 0, it indicates a downward trend; and if Δv = 0, it is considered an unchanged trend or ignored. Continuous data points with the same direction are divided into a single segment. For example, if in a certain periodic segment {v3 = 3.2V, v4 = 3.6V, v5 = 3.9V}, then their Δv values ​​are all greater than 0, and they can be classified as an upward segment. If subsequently {v6 = 3.7V, v7 = 3.2V}, then they can be classified as a downward segment. Data segments within a period can be divided into a set of directional change segments, serving as the structural basis for the trend of the periodic voltage waveform. Each data segment record includes the starting index position, ending position, change direction mark, amplitude start and end values, etc., and can include examples such as "Segment S1 within period P3 is an upward segment, starting and ending points v2 = 2.1V → v6 = 5.3V", generating a waveform directional division set.

[0066] S103: Call the rising and falling segment data from the waveform directional segmentation set, assign interval label numbers to the corresponding segments based on the start and end ranges determined by the positions of the first and last sample points within the segment, and aggregate the classified waveform segments within the period into corresponding sequence label units to generate a periodic waveform segment label set;

[0067] Based on the index positions of the first and last sample points in each segment, start and end range numbers are set. For example, if the first rising segment covers sampling points v3 to v6, the interval number is defined as U1; if it is a falling segment v7 to v11, the number is D1. These numbers are sequentially numbered and recorded to generate an interval identification table. The identified rising / falling segments within a period are aggregated into a complete periodic sequence label unit. For example, if period P1 contains four changing segments {U1, D1, U2, D2}, then the periodic sequence label P1 = {U1, D2} is defined as P1 = {U1, D2}. The tag structure {D1, U2, D2} can be used to describe the voltage trend structure within a period and construct a periodic waveform segment tag set. In practical applications, for example, if the periodic voltage sequence {0.1V, 0.4V, 0.8V, 1.0V, 0.9V, 0.5V, 0.2V} is obtained in a certain operation monitoring, {rising U1: 0.1→1.0V, falling D1: 1.0→0.2V} can be identified, and the corresponding index and number can be recorded to generate a periodic waveform segment tag set.

[0068] Please see Figure 3 The specific steps of S2 are as follows:

[0069] S201: Based on the periodic waveform segmentation marker set, sequentially retrieve the marked rising and falling segment sequences in the periodic segment, extract the connection nodes where the change direction is reversed between adjacent waveform segments, and record the position index of the node in the original periodic segment to generate a waveform reversal node sequence set;

[0070] The system sequentially retrieves the marked ascending and descending segments within each periodic segment. For each adjacent segment pair, it determines whether the direction marker of the current segment (e.g., ascending segment U or descending segment D) is opposite to that of the previous segment. If the direction is reversed (i.e., from ascending to descending, or from descending to ascending), the boundary position of the two segments, i.e., the starting sampling point position of the latter segment, is recorded as a change node. The global index value in the original periodic data sequence is then obtained. For example, in a certain period, if the ending index of segment U1 is n5 and the starting index of segment D1 is n6, then n6 is added as a change node to the reversal point. Node sequence set; In actual implementation, let the voltage sequence be {2.0V, 2.4V, 2.9V, 3.2V, 3.0V, 2.5V, 2.1V, 2.3V}. By partitioning, we can obtain U1: {2.0→3.2V}, D1: {3.2→2.1V}, and U2: {2.1→2.3V}. It can be seen that the reversal nodes appear at the positions from U1→D1 (index 4) and D1→U2 (index 7), respectively. Each node can be accompanied by the numbering record of the adjacent segment pair (such as U1-D1, D1-U2), generating a waveform reversal node sequence set.

[0071] S202: Call the index position between adjacent inverted nodes in the waveform inversion node sequence set, calculate the amplitude slope of the interval by dividing the amplitude change by the corresponding sample point interval within the node interval, arrange them according to the order of nodes, mark the number of inversions of the positive and negative slope between adjacent nodes, and generate a slope change trend sequence group.

[0072] Based on the index positions between any two adjacent nodes, denoted as ni and n{i+1}, extract the original voltage sampling point sequence V={v{n} within the interval. i},v{n i+1},…,v{n{ i+1 Let the starting point be v. start =v{n i The termination point is v. end =v{n{ i+1}}, calculate the voltage amplitude change ΔV=v within the interval. end -v start The number of sample points is Δn = n{ i+1}-n i Then the slope S of the interval can be expressed as S = ΔV / Δn. If ΔV is positive, the slope is positive; if it is negative, the slope is negative. A practical example illustrates this: if n i =10, n{ i+1}=15, v start =3.5V, v end=2.0V, then ΔV=-1.5V, Δn=5, slope S=-0.3V / sampling point, which is negative; after calculating the slope of each adjacent node interval, arrange them in the order of node index, and record whether the sign of the slope of each pair of adjacent intervals changes. For example, if the slope of the previous interval is +0.2V / point and the slope of the next interval is -0.3V / point, it is considered as a direction reversal. Count the number of consecutive reversals of the slope in the whole sequence to generate a slope change trend sequence group.

[0073] S203: Based on the number of node sequences of continuous reversal paths in the slope change trend sequence group, count the total number of nodes with continuous reversal in direction in each reversal path, and compare it with the set slope change identification limit. When the number of nodes exceeds the slope change identification limit, mark the waveform interval within the coverage area of ​​the corresponding path, aggregate the segment index range that meets the conditions, and generate a list of interference candidate change segments.

[0074] Analyze the number of node sequences in each consecutive reversal path. For each path, accumulate the total number of nodes with internal direction reversals, Rcount. If multiple Ti nodes have consecutive direction reversals... If the number of consecutive inverted paths is ≥3, it is considered a continuous inversion path. The number of inverted nodes is counted within the path. A slope change identification threshold Thslope is set, for example, Thslope=3, which means that if the number of consecutive inverted nodes is ≥3, it is judged as an anomaly. The total number of inverted nodes in each path is compared with Thslope. If the condition is met, the original index coverage area corresponding to the path is extracted. For example, if the path {T2, T3, T4} corresponds to the original sampling index segments [50, 60], [60, 70], and [70, 80] respectively, they are merged into the interference candidate mutation segment [50, 80] and added to the interference candidate mutation segment list. If 7 inverted nodes are identified in a certain period of voltage waveform, the corresponding slope direction is... If the number of consecutive reversals is 6, which far exceeds the set threshold of 3, and the corresponding original index range is [100, 160], a list of interference candidate mutation segments is generated.

[0075] Please see Figure 4 The specific steps of S3 are as follows:

[0076] S301: Based on the time node position in the list of candidate mutation segments for interference, configure a fixed-length pre-observation window, and call the original voltage waveform data of the corresponding time period within the observation window. Retrieve the amplitude sign information of continuous alternating positive and negative values ​​in the waveform, determine the zero-crossing intersection position, and record the number of sampling points between adjacent intersections in sequence. After converting to time units, sort them according to the order of appearance to generate a zero-crossing intersection time interval sequence.

[0077] The start time of each abrupt change segment is determined, and a fixed-length pre-observation window is configured beforehand. The length of the observation window can be set to the duration of one or more complete power frequency cycles, such as 40 milliseconds or 60 milliseconds, based on the actual power frequency characteristics of the power equipment. By retrieving the original voltage waveform data corresponding to this time period, the waveform amplitude sequence within the entire window area is searched. The focus is on extracting the positions where the voltage sign changes alternately between positive and negative to identify the zero-crossing points of the voltage waveform. The sign change relationship is judged point by point by traversing the sampling point sequence. Once the signs of two adjacent points are detected to be opposite, a zero-crossing point is considered to exist. The process is continuously executed until all data points within the window are searched. The identified zero-crossing point positions are recorded in sequence, and the number of sampling points between two adjacent points is counted. The sampling frequency is converted into a time unit to obtain the time interval between the points. The points are arranged in the same order as they appear in the waveform to generate a zero-crossing point time interval sequence.

[0078] S302: Call the voltage waveform data within the window corresponding to the zero-crossing intersection time interval sequence, identify the peak and trough points of amplitude according to the extreme value detection method within the zero-crossing period segment, extract the peak amplitude of the period and sort it in time order to generate a peak amplitude difference sequence set;

[0079] The original voltage waveform data within the corresponding observation window is retrieved, and the waveform segment between two adjacent zero-crossing intersections is treated as an independent period segment. The data points within each period segment are traversed to find the maximum and minimum values ​​of the waveform, which are used to identify the peak and trough points of the waveform amplitude. The process must completely cover the waveform segment defined by each pair of zero-crossing intersections. The amplitude of each point is determined by sequential scanning to obtain the maximum and minimum voltage of the period. Then, the peak amplitude of each period is extracted based on the amplitude difference between the two, and the peak amplitudes are sorted according to their chronological order to generate a peak amplitude difference sequence set.

[0080] S303: Using the peak amplitude difference sequence set, call the corresponding zero-crossing intersection time interval sequence of two adjacent observation windows, compare the difference of continuous values ​​in the intersection time interval sequence between two adjacent observation windows, compare whether the change direction of the corresponding position in the peak difference sequence is consistent, and when the continuous change trend is consistent and exceeds the set number of judgments, record the corresponding time period of the window as an abnormal segment and generate a set of abnormal interference stability segments.

[0081] The extracted peak amplitude difference sequence set is combined with the zero-crossing intersection time interval sequence of each of the two adjacent observation windows for analysis. The difference in intersection time interval of each corresponding period segment in the two observation windows is compared in turn to determine whether the change direction is extension or shortening. At the same time, the direction of amplitude change in the position in the peak difference sequence is judged. If the two show a consistent change trend in multiple consecutive periods, it is considered that there is a stable disturbance. The number of times the continuous trend lasts is recorded. When the number of judgments exceeds the preset number, the corresponding time period is marked as an abnormal disturbance stability segment. For example, if the intersection time interval of the two consecutive observation windows both show a shortening and a continuous decrease in peak difference in 3 consecutive periods, it indicates that the voltage waveform has a relatively continuous abnormal change phenomenon in the period. For the time periods that meet the consistent change condition, they are uniformly sorted and numbered to generate a set of abnormal disturbance stability segments.

[0082] Please see Figure 5 The specific steps of S4 are as follows:

[0083] S401: Call the end time point of the cycle of the abnormal segment in the set of abnormal interference stability segments, extend it by a fixed length to construct a voltage recovery monitoring window, extract continuous voltage amplitude sample points in the voltage recovery monitoring window, construct the recovery segment voltage data sequence according to the time order, identify the position of sign change, and generate a recovery segment trend point series dataset.

[0084] Based on the end time point of the abnormal segment as the starting point of the recovery process, a voltage recovery monitoring window is constructed by extending a fixed time range forward. The window length can be consistent with the observation window used in the interference analysis to ensure that the complete voltage recovery process after the interference is covered. Continuous voltage amplitude sample points are extracted within the monitoring window, keeping the sampling interval constant, and a recovery segment voltage data sequence arranged in chronological order is constructed. The data sequence will serve as the basic data for the recovery status, and point-by-point analysis will be performed to identify the positions where the amplitude sign changes from positive to negative, i.e., to identify the zero-crossing points in the voltage waveform. By traversing the sample point sequence, the positive and negative relationships between adjacent samples are detected point by point. Once a sign switch occurs, the point position is recorded as a trend change point. The data is then organized according to the order of occurrence of the change points to generate a recovery segment trend point series dataset.

[0085] S402: Based on the recovery segment trend point series dataset, call the real-time periodic recovery segment and the recovery segment data sequence of the corresponding position of the previous period respectively, and compare the direction node columns of the two in sequence. When the continuous direction change shows an offset trend, locate the position where the trend first changes direction as the starting point, detect the end position of the continuous direction deviation as the ending point, connect the starting point and the ending point to construct a straight line segment, and generate a trend offset line segment sequence.

[0086] The current period's recovery segment trend point list and the previous period's trend point list at the corresponding position are called respectively. The trend change points included in these two sets of data are compared sequentially. A point-to-point method is used to compare the current point with the trend point at the same position in the previous period. The focus is on analyzing whether there is a significant shift in the direction of change between the two points. If there are several consecutive trend points whose direction of change differs from that of the previous period, it can be judged that the trend has shifted. The position where the first difference in direction change occurs is located as the starting point of the shift, and the search continues until the last point where the trend returns to consistency is taken as the ending point. In the waveform data segment from the starting point to the ending point, a linear trend segment is constructed by connecting two points as the descriptive line segment of the trend shift. The process can be repeated until the entire recovery segment data has been traversed, generating a trend shift line segment sequence.

[0087] S403: Call the trend offset line segment sequence and the original reference baseline within the real-time period to determine the point-to-point overlap. Calculate the vertical deviation value from the reference baseline within the coverage area of ​​the trend offset line segment, extract the range where the deviation value continuously exceeds the set deviation threshold, mark the waveform segment corresponding to the deviation exceeding the limit as the correction segment, and obtain the baseline offset correction segment data.

[0088] A point-by-point comparative analysis is performed with the original reference baseline within the real-time period. The overlap of the voltage waveform range covered by each trend offset segment is judged point-to-point. The vertical deviation value between each corresponding point is calculated by comparing the points on the trend segment with the corresponding points on the reference baseline. That is, the offset of the trend point relative to the reference baseline. From all offset values, the part that continuously exceeds the set deviation threshold is selected and recorded in the start and end index interval of the voltage sequence. The complete waveform segment corresponding to the continuous over-limit part is located and separated from the trend offset area as a separate waveform part that needs to be corrected. The segment is the baseline offset correction segment. By processing the correction segment, it can help determine whether the waveform deviates from the normal operating characteristics in certain areas. When slow stabilization or secondary oscillation occurs during the power grid interference recovery process, the marking of the correction segment can effectively help maintenance personnel focus on key recovery areas, improve monitoring accuracy and fault identification capabilities, and obtain baseline offset correction segment data.

[0089] Please see Figure 6 The specific steps of S5 are as follows:

[0090] S501: Based on the time interval corresponding to the baseline offset correction section data, call the original charged judgment baseline data within the real-time detection cycle, determine the starting point of the mapping processing within the correction area as the reference amplitude at the same position in the previous cycle, and locate the endpoint at the offset point amplitude corresponding to the middle position of the real-time recovery segment trend line. Perform linear interpolation between the starting and ending points to construct a transition path and generate a mapping path point series dataset.

[0091] The system retrieves the original charged determination baseline data from the real-time detection cycle, identifies the start and end times of the correction segment, and uses the corresponding position of the time period in the previous cycle as a reference. It extracts the amplitude at the same position in the previous cycle as the correction starting point, locates the offset point of the real-time recovery segment trend line in the middle of the correction segment, and extracts its amplitude as the correction ending point, thus forming an interpolation segment connecting the starting and ending points. Between these two points, linear interpolation is performed according to the chronological order, and the expected amplitude of each point within the interpolation interval is calculated using an equidistant method. A continuous set of transition path point data is constructed point by point. The data points maintain trend consistency with the baseline data of the previous cycle while smoothly transitioning to the offset trend detected in the current cycle, enabling higher-precision reference alignment during fitting and generating a mapped path point dataset.

[0092] S502: Call the interpolation nodes between the start and end points in the mapped path point series dataset, divide it into several continuous sub-segments according to a fixed interval, perform linear fitting processing within the sub-segments, align the nodes within the sub-segments with the original reference baseline, determine the fitting displacement value of each segment through the residual trough value, and generate a reconstructed reference baseline data sequence.

[0093] The system is divided into several continuous sub-segments at fixed intervals, for example, each sub-segment consists of 5 points. Linear fitting is performed on each sub-segment, and the distribution trend of nodes within the segment is analyzed through fitting. The linear fitting results obtained for the sub-segment are compared with the trend of the original reference baseline at the corresponding positions, and the deviation value between the node and the baseline is calculated point by point. The residual trough point in the deviation value is found in each sub-segment as an indicator of representative fitting accuracy. The minimum value is used to determine the optimal fitting displacement value of the segment, and the overall reference position of the sub-segment is adjusted. The fitting results of the sub-segments are concatenated to form a complete fitting adjustment path, ensuring that the new baseline closely follows the trend changes throughout the entire segment, while avoiding excessive response to abrupt noise, and possessing good smoothness and representativeness. This lays an accurate foundation for state correction and re-judgment of conduction status, generating a data sequence for reconstructing the reference baseline.

[0094] S503: Synchronously match the reconstructed reference baseline data sequence with the original conduction state detection record, compare the amplitude of the sampling point with the position of the reconstructed baseline in the detection record, re-mark the comparison result as whether it is in the conduction state or non-conduction state, rewrite the conduction mark sequence according to the judgment state of the sampling point, and generate the compensated energized state correction record.

[0095] The reference baseline data sequence is synchronously matched point by point with the original conduction state detection record. The data value of each sampling point is compared with the amplitude of the reconstructed reference baseline at the same time position. The relative position relationship is used to determine whether it is in a conducting or non-conducting state. If the current sampling value is higher than the reconstructed baseline, it is marked as a conducting state; otherwise, it is marked as a non-conducting state. After point-by-point execution, the original conduction state marking information is updated sequentially to construct a set of conduction judgment results based on the new reference line. After all sampling points are processed, they are used to replace or calibrate the original conduction record sequence. The record can better reflect the misjudgment or omission caused by voltage deviation in actual operation. The compensated state sequence can more accurately identify the conduction and disconnection times, providing a more realistic and reliable data foundation for state switching statistics, electrical equipment operation analysis, and intelligent identification models, and generating a compensated live state correction record.

[0096] The above are merely specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A compensation algorithm for the charged detection result of a power distribution gap lightning arrester, characterized in that, The method comprises the following steps: S1: obtaining a power frequency voltage signal during online operation of a power distribution gap arrester, performing time synchronization processing on the voltage record sequence at a fixed sampling interval, taking the whole cycle voltage zero-crossing point as a cycle division point, classifying the waveform change direction into an ascending section and a descending section, and generating a cycle waveform section marking set; S2: based on the cycle waveform section marking set, extracting a plurality of reverse nodes of the change direction before and after the section, constructing a slope change sequence between the nodes, judging whether the number of reverse trends in the connection path exceeds the slope mutation recognition limit, and generating a disturbance candidate mutation section list; S3: according to the time nodes in the disturbance candidate mutation section list, configuring a fixed-length pre-observation window before the candidate section, extracting the voltage zero-crossing points in the window and recording the interval time, sorting the interval time in time sequence, and generating a disturbance stability abnormal section set; S4: calling the disturbance stability abnormal section set, constructing an extended voltage recovery monitoring window at the end of the cycle section, extracting a recovery section voltage change trend point list, comparing the trend direction of the real-time cycle recovery section with the last cycle recovery section, and obtaining baseline offset correction section data. The cycle waveform section marking set includes waveform amplitude categories, section boundary point positions, and change trend types. The disturbance candidate mutation section list includes mutation start and end times, trend reversal paths, and mutation amplitude characteristics. The disturbance stability abnormal section set includes intersection point spacing variation characteristics, peak value difference distribution, and time variation trend. The baseline offset correction section data includes offset start and end time sequences, trend line fitting parameters, and baseline deviation amplitudes.

2. The compensation algorithm for the live detection result of a power distribution gap surge arrester according to claim 1, characterized in that, The specific steps of S1 are as follows: S101: obtaining a power frequency voltage signal during online operation of a power distribution gap arrester, performing time synchronization processing on the signal sequence according to a fixed sampling interval, identifying a complete power frequency voltage cycle, and combining the sign change of the voltage change direction to detect the amplitude of the voltage sequence point by point. The zero-crossing point position is located in the positive and negative alternating change interval, the zero-crossing point is taken as a division reference point, the voltage signal section with complete cycle structure is divided, and a cycle division sequence set is generated; S102: based on the cycle division sequence set, calling the voltage waveform data in the cycle section, extracting all amplitude sample points in the cycle by equal interval indexing, and judging the direction of the amplitude difference between adjacent sample points in the same cycle. The sample points with the same increasing direction of continuous amplitude are classified into the same change section, and the positive and negative directions of the amplitude difference are classified into ascending sections or descending sections, respectively, to generate a waveform directionality division set; S103: calling the ascending section and descending section data in the waveform directionality division set, assigning interval marker numbers to the corresponding sections according to the start and end ranges determined by the first and last sample point positions in the section, and aggregating the classified waveform sections in the cycle into corresponding sequence marker units to generate a cycle waveform section marking set.

3. The compensation algorithm for the energized detection result of a power distribution gap surge arrester according to claim 2, characterized in that, The specific steps of S2 are as follows: S201: Based on the periodic waveform segment label set, the marked rising segment and the falling segment sequence in the periodic segment are retrieved in sequence, the connection node where the change direction of adjacent waveform segments reverses is extracted, and the position index of the node in the original periodic segment is recorded to generate a waveform inversion node sequence set; S202: The index positions between adjacent inversion nodes in the waveform inversion node sequence set are called, the amplitude slope of the interval is calculated by dividing the amplitude change amount by the interval number of the corresponding sample points within the node interval, the slope is arranged according to the node sequence, the number of inversions of the positive and negative directions of the slope between adjacent nodes is marked, and a slope change trend sequence group is generated; S203: According to the number of node sequences of the continuous inversion path in the slope change trend sequence group, the total number of nodes with continuous direction inversion in each inversion path is counted in sequence, and compared with the set slope mutation recognition limit. When the number of nodes exceeds the slope mutation recognition limit, the waveform interval in the corresponding path coverage range is marked, the index range of the segment meeting the condition is aggregated, and a disturbance candidate mutation segment list is generated.

4. The compensation algorithm for the energized detection result of a power distribution gap surge arrester according to claim 3, characterized in that, The specific steps of S3 are: S301: According to the time node position in the disturbance candidate mutation segment list, a fixed-length pre-observation window is configured, and the original voltage waveform data of the corresponding time segment is called in the observation window. The amplitude symbol information of the waveform with continuous positive and negative alternating changes is retrieved, the zero-crossing point position is determined, and the number of sampling points between adjacent intersection points is recorded in sequence. After conversion to time units, the order is sorted according to the order of appearance, and a zero-crossing intersection time interval sequence is generated; S302: The voltage waveform data in the window corresponding to the zero-crossing intersection time interval sequence is called, the peak value point and the trough point in the zero-crossing period are identified according to the extreme value detection method, and the peak amplitude is extracted and sorted in time sequence to generate a peak amplitude difference sequence set; S303: The peak amplitude difference sequence set is used to call the corresponding zero-crossing intersection time interval sequence in two adjacent observation windows, and the difference between the continuous values in the intersection time interval sequence between the two adjacent observation windows is compared. Whether the change direction of the corresponding position in the peak difference sequence is consistent is compared. When the continuous change trend remains consistent and exceeds the set determination times, the window corresponding time segment is recorded as an abnormal segment to generate a disturbance stability abnormal segment set.

5. The compensation algorithm for the energized detection result of a power distribution gap arrester according to claim 4, characterized in that, The set determination times are not less than three times; The difference comparison between the corresponding zero-crossing intersection time interval sequences in the two adjacent observation windows is limited to the absolute value of the adjacent interval time difference being not less than 1.5 milliseconds; The determination method that the change direction remains consistent is that the change direction of the same position in the continuous three and more peak value difference sequences is the same, and the difference value change direction of the corresponding position in the corresponding intersection time interval sequence is consistent.

6. The compensation algorithm for the energized detection result of a power distribution gap surge arrester according to claim 4, characterized in that, The specific steps of S4 are: S401: The period end time point of the abnormal segment in the disturbance stability abnormal segment set is called, a fixed length is extended to build a voltage recovery monitoring window, and continuous voltage amplitude sample points are extracted in the voltage recovery monitoring window. The recovery segment voltage data sequence is established according to the time sequence, the symbol change position is identified, and a recovery segment trend point column data set is generated; S402: Based on the recovery segment trend point column data set, respectively call the real-time period recovery segment and the corresponding position of the last period recovery segment data sequence, and compare the two direction node columns in sequence. When the continuous direction change appears deviation trend, locate the position where the trend first changes direction as the starting point, detect the end position of the continuous direction deviation as the end point, connect the starting point and the end point to construct a straight line segment, and generate a trend deviation line segment sequence; S403: Call the trend deviation line segment sequence and the original reference baseline in the real-time period for point-to-point coincidence degree judgment. Calculate the vertical deviation value of the reference baseline in the range covered by the trend deviation line segment, and extract the interval range whose deviation value continuously exceeds the set deviation threshold. Mark the waveform segment corresponding to the deviation interval as a correction segment, and obtain the baseline deviation correction section data.

7. The compensation algorithm for the energized detection result of a power distribution gap arrester according to claim 6, characterized in that, The set deviation threshold is not less than 5% of the average amplitude in the recovery segment voltage data sequence; The time interval between the starting point and the end point corresponding to any one of the trend deviation line segments in the trend deviation line segment sequence cannot be less than two complete voltage periods; In the baseline deviation correction section data, only the waveform segment whose deviation value continuously exceeds the set deviation threshold by more than ten sampling points is included.

8. The compensation algorithm for the live detection results of power distribution gap arresters according to claim 1, characterized in that, The algorithm also includes S5 steps: S5: According to the baseline deviation correction section data, perform position mapping processing on the live detection period's live judgment reference line, extract the starting point of the mapping path as the reference value in the last period, and the end point as the middle deviation point in the recovery segment trend line. Perform segmented displacement fitting in the path to reconstruct the reference baseline of the detection period. Re-label the conduction state in the original detection record of the lightning arrester by calling the corrected baseline to generate the compensated live state correction record; The compensated live state correction record includes the corrected baseline trajectory, the conduction state change flag, and the correction timestamp.

9. The compensation algorithm for the energized detection result of a power distribution gap arrester according to claim 8, characterized in that, The specific steps of S5 are: S501: According to the time interval corresponding to the baseline deviation correction section data, call the original live judgment reference line data in the real-time detection period, determine the mapping processing starting point in the correction area as the reference amplitude at the same position in the last period, and the end point is positioned at the deviation point amplitude corresponding to the middle position of the real-time recovery segment trend line. Perform linear interpolation operation between the starting point and the ending point to construct a transition path and generate a mapping path point column data set; S502: Call the interpolation nodes between the starting point and the ending point in the mapping path point column data set, divide them into several continuous sub-sections according to fixed intervals, and perform linear fitting processing in the sub-sections. Align the nodes in the sub-sections with the original reference baseline, determine the fitting displacement value of each section through the residual trough value, and generate a reconstructed reference baseline data sequence; S503: Synchronize the reconstructed reference baseline data sequence with the original conduction state detection record, compare the amplitude of the sampling point with the position relationship of the reconstructed baseline in the detection record, and re-label whether it is in the conduction state or non-conduction state according to the comparison result. Rewrite the conduction mark sequence according to the judgment state of the sampling point, and generate the compensated live state correction record.

Images