Insulation strength detection method of environment-friendly high-voltage electric appliance

Abstract

The invention relates to the technical field of electrical insulation strength measurement, in particular to an insulation strength detection method of an environment-friendly high-voltage electrical appliance, which comprises the following steps of: acquiring temperature and pressure data, calculating a first-order derivative identification response and stress point recording phase difference, generating a time sequence window, constructing a phase difference sequence, marking a phase deviation interval through Mann-Kendall inspection, and calculating a phase difference value; and extracting a corresponding leakage current abnormal peak value to mark a local abnormal interval, analyzing a resistance strain peak value and a relaxation trend to mark a fatigue window, comparing a phase and current abnormal interval to detect an insulation risk node, and screening trend consistency and stable fluctuation nodes to generate an insulation strength confidence critical node set. According to the invention, temperature and voltage data are collected synchronously, derivative and phase difference are calculated to identify a sudden change point, sensitivity is improved through cooperative trend identification, current strain rhythm is fused to capture a fatigue area, a risk area is positioned through judgment of signal superposition, and insulation strength early warning precision is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of insulation strength measurement, and in particular to an insulation strength detection method for an environmentally friendly high-voltage electrical appliance. Background Art

[0002] The field of insulation strength measurement technology encompasses the evaluation and testing of the insulation performance of electrical equipment, with key applications in the power, electronics, and communications industries. The core focus is to test the insulation capabilities of electrical equipment in high-voltage environments through various means to ensure that the equipment does not experience electrical failures during operation. Insulation strength testing not only assesses equipment safety but also effectively predicts the risk of long-term failure. Key equipment involved includes transformers, switches, and line cables. These tests are typically conducted using DC or AC high-voltage test equipment in conjunction with electrical insulation performance standards.

[0003] The insulation strength testing method for environmentally friendly high-voltage electrical appliances involves measuring the insulation strength of high-voltage electrical appliances using specific testing methods and equipment to assess their safety and stability during actual operation. To address the issue of testing the insulation layer of high-voltage electrical appliances, a combination of current and voltage measurements is used to determine insulation performance. By conducting detailed tests on the insulating materials and structures of electrical equipment, its insulation strength can be assessed without damaging the equipment. This involves equipment such as a high-voltage current source, test electrodes, and measuring instruments, aiming to improve the precision and reliability of the testing process and ensure the accuracy of test results.

[0004] Existing technologies for insulation strength testing generally rely on steady-state testing of single-channel electrical parameters, making them inadequate for the dynamic, interactive environment in which equipment operates. Traditional methods typically rely solely on static voltage or current response data, failing to capture subtle trends in signal coupling, leading to blind spots in detecting early stages of insulation degradation. The lack of a mechanism for comparing multiple physical quantities over a continuous time period prevents in-depth analysis from the perspective of signal rhythm or structural changes, neglecting the potential temporal co-evolution of signals such as temperature, voltage, and strain. For example, traditional leakage current detection methods are often affected by short-term electromagnetic interference, resulting in poor peak identification stability and a coexistence of false positives and missed detections. Furthermore, without the ability to incorporate mechanical monitoring, there is a lack of supplementary assessment methods for material fatigue behavior, making it impossible to assess the risk of insulation degradation after equipment is subjected to operational stress. These deficiencies impact the accuracy of safety assessments and the efficiency of operational and maintenance decisions in complex operating environments. Summary of the Invention

[0005] The purpose of the present invention is to solve the shortcomings of the prior art and to propose an environmentally friendly high-voltage electrical insulation strength detection method.

[0006] In order to achieve the above-mentioned purpose, the present invention adopts the following technical solution: a method for detecting the insulation strength of an environmentally friendly high-voltage electrical appliance, comprising the following steps: S1: Collect temperature and voltage data of high-voltage electrical appliances, calculate the first-order derivative within the sliding window, identify the point where the temperature derivative changes sign as the response trigger point, extract the voltage derivative mutation point as the stress excitation point, calculate the temperature and pressure phase difference between the two in the same window, record the window number, and generate a time series window sequence; S2: Based on the time series window sequence, the temperature and pressure phase difference between consecutive windows is calculated and a first-order difference sequence is constructed. The direction consistency is identified and the offset interval is marked through the Mann-Kendall trend test to generate a phase offset interval. S3: acquiring a leakage current signal of a corresponding period based on the phase offset interval, detecting a current peak value and analyzing a sliding trend of adjacent peak points, determining reverse fluctuation behavior and marking abnormal peak values, and generating a local abnormal peak interval; S4: Collect resistance strain data using a resistance strain gauge, extract the resistance strain response peak and relaxation duration using a sliding window, determine whether there is a coordinated trend of peak increase and relaxation shortening using the STL algorithm, and mark potential fatigue windows. Cross-time comparison is performed on the phase offset interval and the local abnormal peak interval to detect critical risk nodes of insulation strength; S5: Based on the insulation strength critical risk nodes, the k-nearest neighbor algorithm is used to judge the trend consistency of the overlapping windows, and the node groups with trend consistency and smooth fluctuation in the front and rear neighborhood windows are screened to generate an insulation strength confidence critical node set.

[0007] As a further solution of the present invention, the timing window sequence is specifically the temperature derivative sign change point number, the voltage derivative mutation point number, and the temperature-pressure phase difference. The phase offset interval includes the offset start time, the offset end time, and the phase difference variation amplitude. The local abnormal peak interval includes the abnormal peak time number, the abnormal peak amplitude, and the adjacent peak fluctuation direction. The insulation strength critical risk node is specifically the risk node time number, the risk node insulation strength, and the risk node corresponding window number. The insulation strength confidence critical node set includes the node time number, the stable fluctuation trend node group, and the node trend consistency number group.

[0008] As a further solution of the present invention, the specific steps of S1 are: S101: Collect temperature and voltage data of high-voltage electrical appliances, traverse the data sequence using a sliding window, calculate the first-order derivatives of adjacent points in the temperature sequence, determine the locations of the derivative sign change points based on the sign switching between adjacent derivatives, summarize them, and generate a set of temperature derivative sign change locations; The high-voltage electrical appliance is a high-voltage circuit breaker; S102: Based on the temperature derivative sign change position set, voltage data within each window is acquired, a voltage first-order derivative value sequence is calculated, a time point at which the derivative mutation amplitude exceeds a voltage mutation recognition threshold is selected as a voltage mutation point, and the time point is matched with the temperature sign change point to generate a voltage mutation point position set; The voltage mutation recognition threshold is set by calculating the mean and standard deviation of the voltage first-order derivative change amplitude sequence within the statistical window and combining it with the sensitivity adjustment coefficient k; S103: Call the timestamps of the temperature derivative sign change position set and the voltage mutation point position set, calculate each pair of time differences, convert the time differences into phase angle values ​​in combination with the current sliding window length, record the window number, and generate a time series window sequence.

[0009] As a further solution of the present invention, the specific steps of S2 are: S201: Calculating a phase difference sequence between two adjacent windows based on the time series window sequence, and processing data corresponding to all window numbers to obtain a phase difference change sequence; S202: calling the phase difference change value sequence, calculating the positive sequence statistic value of the window in the sequence by the Mann-Kendall trend test method, and calculating the reverse sequence statistic value at the same time, and determining the direction consistency interval according to the cross relationship between the positive sequence statistic value and the reverse sequence statistic value, and obtaining the trend direction consistent number set; S203: According to the trend direction consistent number set, backtrack to obtain the start and end time segments of the corresponding number window in the original time series window phase difference sequence, combine the boundary numbers of the number interval marked with the phase difference direction consistency information in the time segment, and generate a phase offset interval.

[0010] As a further solution of the present invention, the positive sequence statistic value of the window in the sequence is calculated by the Mann-Kendall trend test method. , using the formula: ; in, Represents the phase difference change value sequence The phase difference change of each element is Represents the window Elements to The inverse weight factor of the window spacing, Represents the window The element and The time attenuation coefficient of the device thermal inertia of the element, n represents the number of phase difference change values ​​included in the current window, and sgn represents the sign function.

[0011] As a further solution of the present invention, the specific steps of S3 are: S301: acquiring the leakage current original signal of the corresponding time period according to the phase offset interval, extracting the current at multiple time points and detecting the positions of local maximum points, marking them as current peak points, and generating a leakage current peak point set; S302: calling the leakage current peak point set, sequentially comparing the current magnitudes of the current peak point with the previous peak point, calculating a trend change identification sequence indicating whether the change direction is rising or falling, and analyzing the occurrence positions of continuous direction reversals in the sequence to generate a reverse sliding trend identification set; S303: extracting the corresponding current peak point number according to the reverse sliding trend identification set, and tracing back the original peak point position and time series number information, marking the time number interval of the peak reversal point in each continuous reverse sliding interval, and generating a local abnormal peak interval.

[0012] As a further solution of the present invention, the specific steps of S4 are: S401: Collect resistance strain data at multiple time points on the surface of the insulating medium using a resistance strain gauge, calculate the resistance strain change rate within a sliding window, and extract the response wave peak value corresponding to the window and the relaxation time corresponding to the strain decaying to a stable state after the peak value, to obtain a resistance strain characteristic sequence; S402: Based on the resistance strain characteristic sequence, calculate the local trend component of the relaxation time using the STL algorithm, and determine whether there are synchronous segments in which the peak value shows an upward trend and the relaxation time shows a shortening trend in multiple consecutive windows, to obtain the fatigue trend synchronization interval; S403: Perform cross time period judgment based on the fatigue trend synchronization interval, the phase offset interval, and the local abnormal peak interval, filter a set of time nodes that exist in the time period at the same time, and obtain an insulation strength critical risk node.

[0013] As a further solution of the present invention, the local trend component of the relaxation time is calculated by the STL algorithm , using the formula: ; in, represents the ath relaxation duration sampling value of the wth window, represents the arithmetic mean of the relaxation time of the w-th window, represents the bth resistance strain characteristic peak in the wth window, represents the total duration of the w-th time window, The equipment characteristic coefficients determined by calibration experiments, , represents the window time span of the w-th window, Represents the baseline relaxation time constant.

[0014] As a further solution of the present invention, the specific steps of S5 are: S501: Based on the insulation strength critical risk node, obtain the resistance strain peak change trend within the overlapping time window and the change direction of the adjacent window, classify them, calculate the Euclidean distance between the current node and the neighboring nodes using the k-nearest neighbor algorithm, and classify them to obtain a trend consistency matching node set; S502: calling the trend consistency matching node set, calculating the difference between the maximum and minimum values ​​of the resistance strain in the time series corresponding to each group of nodes as the fluctuation amplitude, and filtering the node group by the stability threshold to obtain a stable fluctuation trend node group; The stability threshold is set by the sum of the mean and standard deviation of the resistance strain amplitude in all cross windows; S503: extracting index positions in the time series according to the stable fluctuation trend node group, and performing set mapping processing on the index time points to generate an insulation strength confidence critical node set.

[0015] Compared with the prior art, the advantages and positive effects of the present invention are: In the present invention, by synchronously collecting temperature and voltage data of environmentally friendly high-voltage electrical appliances, high-precision identification of temperature mutation points and voltage fluctuation points is performed based on first-order derivative calculation and phase difference construction within a sliding window, thereby improving the ability to grasp the interaction patterns of multi-source signals in the operating environment. The first-order difference processing of phase difference and the trend consistency identification strategy make the identification of slow changes in the internal electrical performance of the equipment more sensitive, thereby capturing the dynamic trend of early offset. The peak extraction and reverse fluctuation detection process of the leakage current introduces a sliding trend identification mechanism, effectively avoiding misjudgments caused by occasional interference and enhancing the positioning accuracy of local anomalies. Further combining the synergistic trend between the strain response peak and the relaxation time, the potential fatigue interval is captured through rhythm changes, reflecting the systematic control of the stress changes of the insulating material and the coupling behavior of the electrical signal. The abnormal overlapping sections of multiple signal sources in the same time window are cross-compared to perform composite trigger identification of critical insulation strength risks, improving the foresight of the prediction and the pertinence of abnormal event detection. The entire process breaks the limitations of monitoring a single physical quantity, strengthens the collaborative analysis capabilities of multi-dimensional information, enhances the capture accuracy in the early stages of insulation state evolution, and provides predictive warnings of safety margins for equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 It is a schematic diagram of the main steps of the present invention. DETAILED DESCRIPTION

[0017] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0018] In the description of the present invention, it should be understood that the terms "length," "width," "up," "down," "front," "back," "left," "right," "vertical," "horizontal," "top," "bottom," "inside," "outside," and the like, indicating positions or relationships, are based on the positions or relationships shown in the accompanying drawings and are intended only to facilitate the description of the present invention and simplify the description. They do not indicate or imply that the devices or elements referred to must have a specific orientation, be constructed, or operate in a specific orientation. Therefore, they should not be construed as limiting the present invention. Furthermore, in the description of the present invention, "plurality" means two or more, unless otherwise expressly and specifically defined.

[0019] See also Figure 1 The present invention provides a technical solution: a method for detecting the insulation strength of an environmentally friendly high-voltage electrical appliance, comprising the following steps: S1: Collect temperature and voltage data of high-voltage electrical appliances, calculate the first-order derivative within the sliding window, identify the point where the temperature derivative changes sign as the response trigger point, extract the voltage derivative mutation point as the stress excitation point, calculate the temperature and pressure phase difference between the two in the same window, record the window number, and generate a time series window sequence; S2: Based on the time series window sequence, the temperature and pressure phase difference between consecutive windows is calculated and the first-order difference sequence is constructed. The direction consistency is identified and the offset interval is marked through the Mann-Kendall trend test to generate the phase offset interval; S3: Acquire the leakage current signal of the corresponding period based on the phase offset interval, detect the current peak and analyze the sliding trend of adjacent peak points, determine the reverse fluctuation behavior and mark the abnormal peak, and generate the local abnormal peak interval; S4: Collect resistance strain data using a resistance strain gauge, extract the resistance strain response peak and relaxation duration using a sliding window, use the STL algorithm to determine whether there is a synergistic trend of peak increase and relaxation shortening and mark potential fatigue windows, perform a cross-time comparison between the combined phase offset interval and the local abnormal peak interval to detect critical risk nodes of insulation strength; S5: Based on the insulation strength critical risk nodes, the k-nearest neighbor algorithm is used to judge the trend consistency of the overlapping windows, and the node groups with consistent trends and smooth fluctuations in the front and rear neighborhood windows are screened to generate the insulation strength confidence critical node set.

[0020] The time series window sequence specifically includes the temperature derivative sign change point number, the voltage derivative mutation point number, and the temperature-pressure phase difference. The phase offset interval includes the offset start time, the offset end time, and the phase difference variation amplitude. The local abnormal peak interval includes the abnormal peak time number, the abnormal peak amplitude, and the adjacent peak fluctuation direction. The insulation strength critical risk node specifically includes the risk node time number, the risk node insulation strength, and the risk node corresponding window number. The insulation strength confidence critical node set includes the node time number, the stable fluctuation trend node group, and the node trend consistency number group.

[0021] See also Figure 1 , the specific steps for obtaining S1 are: S101: Collect temperature and voltage data of high-voltage electrical appliances, set a sliding window length and traverse the data sequence, calculate the first-order derivatives of adjacent points in the temperature sequence, determine the locations of the derivative sign change points based on the sign switching between adjacent derivatives, summarize them, and generate a set of temperature derivative sign change locations; The high-voltage electrical appliance is a high-voltage circuit breaker; Environmentally friendly high-voltage circuit breakers are electrical switching devices used in high-voltage power transmission systems. They have the ability to quickly cut off current when a circuit fault occurs. Their environmentally friendly features are mainly reflected in the green improvements in the insulating medium and structural design. Traditional high-voltage circuit breakers mostly use sulfur hexafluoride gas as an insulating and arc-extinguishing medium. Although it has excellent performance, it has an extremely high greenhouse effect potential. Environmentally friendly high-voltage circuit breakers use alternative gases (such as dry air, nitrogen, or new mixed gases) to effectively reduce greenhouse gas emissions. At the same time, in terms of material selection, they give priority to the use of recyclable, low-carbon footprint insulation materials. The entire design cycle considers equipment energy efficiency, carbon emissions during the manufacturing process, and recycling and disposal after the end of its service life, reflecting comprehensive control of the impact on the ecological environment. The temperature data sequence [25.0, 25.2, 25.5, 25.3, 25.1, 24.9, 24.8] of a high-voltage circuit breaker is collected with a window length of T = 5 seconds. The derivative sequence of the temperature difference between adjacent points is calculated as [0.2, 0.3, -0.2, -0.2, -0.2, -0.1]. When traversing the derivative sequence, the derivative 0.2 at index 1 and 0.3 at index 2 have the same sign, while the derivative 0.3 at index 2 and -0.2 at index 3 have different signs. Index 2 is recorded as the sign change point, and index 3 is recorded as -0. 2 has the same sign as -0.2 at index 4, -0.2 at index 4 has the same sign as -0.2 at index 5, and -0.2 at index 5 has the same sign as -0.1 at index 6, generating a sign change point position set [2]. As shown in Table 1, the temperature derivative sign change point identification process, when the window slides to the 3rd second, the temperature sequence becomes [25.5, 25.3, 25.1, 24.9, 24.8], and the derivative sequence is calculated as [-0.2, -0.2, -0.2, -0.1]. No sign change is detected, and the sign change point set remains empty.

[0022] Table 1 Example table of identification of temperature derivative sign change points

[0023] S102: Based on the temperature derivative sign change position set, voltage data within each window is obtained, a voltage first-order derivative value sequence is calculated, and a time point at which the derivative mutation amplitude exceeds the voltage mutation recognition threshold is selected as a voltage mutation point and matched with the temperature sign change point to generate a voltage mutation point position set; The voltage mutation recognition threshold is set by calculating the mean and standard deviation of the voltage first-order derivative change amplitude sequence within the statistical window and combining it with the sensitivity adjustment coefficient k; Take the voltage data sequence in the window [10.0, 10.2, 10.5, 10.3, 10.1], calculate the voltage derivative value sequence [0.2, 0.3, -0.2, -0.2], and get the average value of the derivative. , standard deviation , set the sensitivity adjustment coefficient k=2, and calculate the voltage mutation recognition threshold as , traversing the derivative value sequence, 0.2<0.525 at index 1 is not marked, 0.3<0.525 at index 2 is not marked, and |-0.2|=0.2<0.525 at index 3 is not marked. When the window slides to the 4th second, the voltage sequence is [10.5, 10.3, 10.1, 9.8, 9.5], and the derivative value sequence is [-0.2, -0.2, -0.3, -0.3]. The maximum mutation amplitude of 0.3 is still lower than the threshold of 0.525, and no effective mutation point is detected.

[0024] S103: Calling the timestamps of the temperature derivative sign change position set and the voltage mutation point position set, calculating the time difference of each pair, converting the time difference into a phase angle value based on the current sliding window length, and recording the window number to generate a time series window sequence; Take the temperature sign change point timestamp t1 = 2.0s and the voltage mutation point timestamp t2 = 2.1s, calculate the time difference Δt = 0.1s, set the window length T = 5s, convert the phase angle to (0.1 / 5) × 360 = 7.2°, record the window number W001 and the phase value 7.2°. When processing the third window, the time difference between the temperature sign change point t3 = 5.2s and the voltage mutation point t4 = 5.3s is 0.1s. The phase angle is converted to (0.1 / 5) × 360 = 7.2°. Generate the time series window sequence [W001: 7.2°, W003: 7.2°] and perform the conversion. When the time difference reaches 0.15s, the phase angle is converted to (0.15 / 5) × 360 = 10.8°, which exceeds the preset phase tolerance range and triggers an alarm.

[0025] See also Figure 1 , the specific steps for obtaining S2 are: S201: Calculate a phase difference sequence between two adjacent windows based on the time series window sequence, and process data corresponding to all window numbers to obtain a phase difference change sequence; Take the time series window sequence [W001: 7.2°, W002: 0.0°, W003: 7.2°], extract the phase angle value [7.2, 0.0, 7.2], calculate the phase difference of adjacent windows Δφ1=0.0-7.2=-7.2°, Δφ2=7.2-0.0=7.2°, generate the difference sequence [-7.2, 7.2], when processing the fourth window W004 phase angle of 10.8°, calculate Δφ3=10.8-7.2=3.6°, the difference sequence is expanded to [-7.2, 7.2, 3.6]. When traversing all window numbers, if the window interval is 5 seconds, the interval between windows W001 and W002 is 5 seconds, and the interval between W002 and W003 is 10 seconds. The difference sequence is recorded in chronological order as [-7.2, 7.2, 3.6]. As shown in Table 2, the phase difference change value calculation process, when the phase difference of a single window is -3.6°, the difference with the previous difference of 7.2° is -10.8°, and the sequence is updated to [-7.2, 7.2, 3.6, -10.8].

[0026] Table 2 Phase difference change value calculation example table

[0027] S202: calling the phase difference change value sequence, calculating the positive sequence statistic value of the window in the sequence by the Mann-Kendall trend test method, and calculating the reverse sequence statistic value at the same time, and determining the direction consistency interval based on the cross relationship between the positive sequence statistic value and the reverse sequence statistic value, and obtaining the trend direction consistent number set; Take the difference sequence [-7.2, 7.2, 3.6], Representative The phase difference change of each window is obtained by subtracting the phase angles of adjacent windows, such as the phase angle of window W002 Phase angle with W001 The difference is , Represents the inverse weight factor of the window spacing, and the calculation formula is , the distance between windows W001 and W003 is 2 units ( ),but , Represents the thermal inertia time decay coefficient of the equipment, which is set according to the experimental data of the circuit breaker heat dissipation characteristics. , when the window spacing is 2 units , represents the symbolic function, Output 1 when Output 1 when Output 0 when the phase difference changes , window number [W001, W002, W003], calculate and The item value represents; 1. Time Representative , , , , , substitute into the formula: ; 2. Time represents: , , , , , , 3. Accumulated result represents , when calculating the reverse statistic, j=2, i=1, the value is -2.08, j=3, i=1, the value is -3.95, the cumulative value is ,When the ratio of the positive order statistic 6.03 to the reverse order statistic -6.03 exceeds the threshold 1.5, the window [W001, W003] is determined to be a direction-consistent interval.

[0028] S203: Based on the trend direction consistent number set, backtrack to obtain the start and end time segments of the corresponding numbered window in the original time series window phase difference sequence, combine the boundary numbers of the numbered intervals with the phase difference direction consistency information within the time segment, and generate the phase offset interval; Based on the trend consistent number set [W001, W003], the original phase difference sequence timestamps [2.0s, 5.2s] are traced back to calculate the time segment length of 5.2-2.0=3.2 seconds, marking the start number W001 (2.0s) and the end number W003 (5.2s). When it is detected that window W004 (10.8°) is not included in the consistent interval, the boundary remains at [W001, W003]. If the phase difference change value of window W005 is -5.4°, resulting in a trend reversal, the number set is updated to [W001, W003, W005], and the time segment is extended to 5.2s to 15.6s, generating the phase offset interval [W001: 2.0s, W005: 15.6s].

[0029] See also Figure 1 , the specific steps for obtaining S3 are: S301: Obtaining the leakage current original signal of the corresponding time period according to the phase offset interval, extracting the current at multiple time points and detecting the location of local maximum points, marking them as current peak points, and generating a leakage current peak point set; Get the leakage current data sequence [1.2, 1.5, 1.8, 2.1, 1.9, 1.7] mA for the time period corresponding to the phase offset interval [W001: 2.0s, W003: 5.2s]. Extract the current 1.2 mA at time points t=2.0s, 1.5 mA at time points t=2.5s, and 1.8 mA at time points t=3.0s at 0.5 second intervals. When traversing the current sequence, the current 1.5 mA at index 2 is compared with the adjacent 1.2 mA at index 1 and 1.8 mA at index 3. Since 1.5>1.2 and 1.5<1.8 are not marked The peak value is 1.8mA at index 3, which satisfies 1.8>1.5 and 1.8<2.1, so it is not marked. The peak value 2.1mA at index 4 satisfies 2.1>1.8 and 2.1>1.9, so it is marked as peak point P001. The peak value 1.9mA at index 5 is not marked because 1.9<2.1, so the peak point set [P001: 2.1mA.4.0s] is generated. As shown in Table 3, the current peak detection process is as follows: when the window is expanded to 5.2s, the current sequence is [1.9, 1.7, 2.0, 2.3, 2.1], and the peak point 2.3mA at index 3 is detected as the peak point P002.

[0030] Table 3 Leakage current peak detection example

[0031] S302: Calling the leakage current peak point set, comparing the current magnitudes of the current peak point with the previous peak point in sequence, calculating a trend change identification sequence indicating whether the change direction is rising or falling, and analyzing the occurrence positions of continuous direction reversals in the sequence to generate a reverse sliding trend identification set; Take the peak point set [P001: 2.1mA, P002: 2.3mA, P003: 2.1mA], calculate the current change ΔI1 between P002 and P001 = 2.3-2.1 = 0.2mA > 0 and mark it as an upward trend ↑, the current change ΔI2 between P003 and P002 = 2.1-2.3 = -0.2mA < 0 and mark it as a downward trend ↓, and generate a trend sequence [↑,↓]. When continuous trend changes such as [↑,↓,↑] are detected, the ↓ at index 2 and the ↑ at index 3 constitute a direction reversal, and the reversal position index 2 is recorded. If the trend sequence [↓,↑,↓] appears, the reversal is recorded at indexes 1 and 3, and the reverse sliding identification set [2,3] is generated. The trend reversal detection process is shown as follows. When the trend sequence is [↑,↑,↓], only one reversal is recorded at index 2.

[0032] S303: extracting the corresponding current peak point number according to the reverse sliding trend identification set, and retracing the original peak point position and time series number information, marking the time number interval of the peak reversal point in each continuous reverse sliding interval, and generating a local abnormal peak interval; According to the reverse sliding identification set [2, 3], corresponding to the peak points P002 (5.2s) and P003 (6.5s), the original time series number [W003, W004] is traced back to calculate the time interval 6.5-5.2=1.3 seconds, marking the start number W003 (5.2s) and the end number W004 (6.5s). When a new reversal point W005 (8.0s) is detected, the interval is expanded to [W003: 5.2s, W005: 8.0s]. If the interval includes three consecutive reversal points, it is split into [W003: 5.2s, W004: 6.5s] and [W004: 6.5s, W005: 8.0s] to generate the local abnormal peak interval [W003-W005].

[0033] See also Figure 1 , the specific steps for obtaining S4 are: S401: Collect resistance strain data at multiple time points on the surface of the insulating medium using a resistance strain gauge, calculate the resistance strain change rate within a sliding window, and extract the response wave peak value corresponding to the window and the relaxation time corresponding to the strain decaying to a stable state after the peak value, to obtain a resistance strain characteristic sequence; The resistance strain data sequence [120, 125, 130, 128, 125, 122] with an interval of 0.2 seconds was collected, and the window length T was set to 1 second (5 data points). The strain change rate was calculated as (122 + 120) / (0.8) = 2.5. The peak value 128 (index 3) in the detection window was detected, and the data after the peak value [128, 125, 122] was recorded. The time required for decay to a stable state (fluctuation < 1) was calculated to be 0.6 seconds. When the window slid to the second second, the data [125, 130, 135, 132, 128] had a peak value of 135 and a decay time of 0.8 seconds. The feature sequence [(128, 0.6), (135, 0.8)] was generated. The feature extraction process is shown in Table 4. When a secondary rise (including data [122, 125, 123]) appeared after the peak value of a single window, the decay time was recalculated.

[0034] Table 4 Example table of resistance strain feature extraction

[0035] S402: Based on the resistance strain characteristic sequence, the local trend component of the relaxation time is calculated using the STL algorithm, and it is determined whether there are synchronous segments in which the peak value shows an upward trend and the relaxation time shows a shortening trend in multiple consecutive windows, thereby obtaining the fatigue trend synchronization interval; Take the feature sequence [(128, 0.6), (135, 0.8), (140, 1.2)], Representative Window No. The relaxation time sampling value is obtained by measuring the time it takes for the strain to decay to a stable state (fluctuation < 1) after the peak, including the three sampling values ​​in window W001 , Representative The arithmetic mean of the relaxation time of the windows is calculated as follows: ,when and hour, , Representative Window No. The characteristic peak value of resistance strain is directly taken from the monitoring data, the peak sequence of window W002 , Represents the total window duration, fixed at 1.0 seconds (5 data points, sampling interval 0.2 seconds), representing Represents the equipment characteristic coefficient, determined through calibration experiments, measured when 80% of the rated load is applied to a single model circuit breaker , in line with the normal range , Represents the window time span, and the sliding window interval is 1.0 second, so , represents the baseline relaxation time constant, taking the median of normal data as 1.0 second; Example calculation, take window W002 parameter representative, molecular calculation representative , At the same time, remove the denominator to calculate the representative , , , , threshold determination is associated with the result, setting the fatigue trend threshold ,because , indicating that the combination of the dispersion of relaxation duration and peak intensity in window W002 does not meet the fatigue accumulation standard. When the peak value increases (128→135→140) and the relaxation duration decreases (0.6→0.8→1.2 in reverse) in three consecutive windows, it is determined to be an abnormal trend and the synchronization interval [W001, W003] is generated.

[0036] S403: Based on the fatigue trend synchronization interval, the phase offset interval, and the local abnormal peak interval, cross-time period judgment is performed, and a set of time nodes that exist in the time period are screened to obtain the insulation strength critical risk node; According to the fatigue trend synchronization interval [W001, W003: 2.0-4.0s], the time intersection judgment is performed with the phase offset interval [W003, W005: 5.2-8.0s] and the abnormal peak interval [W003, W005: 5.2-8.0s]. There is no overlap in the 2.0-4.0s interval, and there is complete overlap in the 5.2-8.0s interval. The key nodes [5.2, 6.5, 8.0]s in the overlap interval are extracted based on the data sampling points. When the nodes simultaneously meet the local trend component When the phase angle is greater than 5°, the node 5.2s that does not meet the conditions is eliminated and the critical risk nodes [6.5s, 8.0s] are marked.

[0037] See also Figure 1 , the specific steps for obtaining S5 are: S501: Based on the insulation strength critical risk node, the resistance strain peak change trend within the overlapping time window and the change direction of the adjacent window are obtained and classified. The Euclidean distance between the current node and the neighboring nodes is calculated using the k-nearest neighbor algorithm and classified to obtain a trend consistency matching node set; Extract the resistance strain peak sequence [142, 145, 148] of the window W004-W006 corresponding to the critical risk node [6.5s, 8.0s], calculate the change trend of adjacent windows: W004→W005 window change trend Δ1=145-142=3 (increasing), W005→W006 window change trend Δ2=148-145=3 (increasing), classified as a continuous rising trend class, set k=3 nearest neighbors, select three windows before and after the node 8.0s [W003: 140, W004: 142, W005: 145], calculate the Euclidean distance: node 8.0s feature vector [148, 3] (peak, change), W003 distance: , W004 distance: , W005 distance: , the nearest neighbors are W005 (3.0), W004 (6.0), and W003 (8.25), all of which belong to the rising trend class, generating the matching node set [W003-W005].

[0038] Table 5 Example of Euclidean distance calculation

[0039] S502: calling the trend consistency matching node set, calculating the difference between the maximum and minimum values ​​of the resistance strain in the time series corresponding to each group of nodes as the fluctuation amplitude, and filtering the node group by the stability threshold to obtain the stable fluctuation trend node group; The stability threshold is set by the sum of the mean and standard deviation of the resistance strain amplitude in all cross-windows; Extract the matching node set [W003: 140, W004: 142, W005: 145], calculate the fluctuation amplitude of each window: W003 amplitude = 140-135 = 5με, W004 amplitude = 142-138 = 4με, W005 amplitude = 145-140 = 5με, and calculate the global amplitude mean , standard deviation , set the stability threshold , screen the windows with amplitude ≤ 5.14 [W003: 5, W004: 4, W005: 5] to generate a stable fluctuation group.

[0040] S503: extracting index positions in the time series according to the stable fluctuation trend node group, and performing set mapping processing on the index time points to generate an insulation strength confidence critical node set; The stable group window time index [W003: 5.2s, W004: 6.5s, W005: 8.0s] is extracted, and a timestamp mapping table is established: 5.2s→index 101, 6.5s→index 130, 8.0s→index 160. When the amplitude of the new window W006 (9.5s) is 6 and exceeds the threshold 5.14, it is excluded, and the insulation strength confidence critical node set [101, 130, 160] is generated.

[0041] The above are merely preferred embodiments of the present invention and do not limit the present invention in any other form. Any technician familiar with the profession may use the technical content disclosed above to change or modify it into an equivalent embodiment with equivalent changes and apply it to other fields. However, any simple modification, equivalent change and modification made to the above embodiment based on the technical essence of the present invention without departing from the content of the technical solution of the present invention shall still fall within the scope of protection of the technical solution of the present invention.

Claims

1. A method for detecting the insulation strength of an environmentally friendly high-voltage electrical appliance, characterized in that: The following steps are involved: S1: Collect temperature and voltage data of high-voltage electrical appliances, calculate the first-order derivative within the sliding window, identify the point where the temperature derivative changes sign as the response trigger point, extract the voltage derivative mutation point as the stress excitation point, calculate the temperature and pressure phase difference between the two in the same window, record the window number, and generate a time series window sequence; S2: Based on the time series window sequence, the temperature and pressure phase difference between consecutive windows is calculated and a first-order difference sequence is constructed. The direction consistency is identified and the offset interval is marked through the Mann-Kendall trend test to generate a phase offset interval. S3: acquiring a leakage current signal of a corresponding period based on the phase offset interval, detecting a current peak value and analyzing a sliding trend of adjacent peak points, determining reverse fluctuation behavior and marking abnormal peak values, and generating a local abnormal peak interval; S4: Collect resistance strain data through a resistance strain gauge, extract the resistance strain response peak and relaxation time through a sliding window, determine whether there is a coordinated trend of peak rise and relaxation shortening through the STL algorithm and mark the potential fatigue window, combine the phase offset interval and the local abnormal peak interval for cross-time comparison, and detect the critical risk node of insulation strength.

2. The insulation strength testing method of an environmentally friendly high-voltage electrical appliance according to claim 1, characterized in that: The timing window sequence specifically includes the temperature derivative sign change point number, the voltage derivative mutation point number, and the temperature-pressure phase difference. The phase offset interval includes the offset start time, the offset end time, and the phase difference variation amplitude. The local abnormal peak interval includes the abnormal peak time number, the abnormal peak amplitude, and the adjacent peak fluctuation direction. The insulation strength critical risk node specifically includes the risk node time number, the risk node insulation strength, and the risk node corresponding window number.

3. The insulation strength testing method of an environmentally friendly high-voltage electrical appliance according to claim 1, characterized in that: The specific steps of S1 are: S101: Collect temperature and voltage data of high-voltage electrical appliances, traverse the data sequence using a sliding window, calculate the first-order derivatives of adjacent points in the temperature sequence, determine the locations of the derivative sign change points based on the sign switching between adjacent derivatives, summarize them, and generate a set of temperature derivative sign change locations; The high-voltage electrical appliance is a high-voltage circuit breaker; S102: Based on the temperature derivative sign change position set, voltage data within each window is acquired, a voltage first-order derivative value sequence is calculated, a time point at which the derivative mutation amplitude exceeds a voltage mutation recognition threshold is selected as a voltage mutation point, and the time point is matched with the temperature sign change point to generate a voltage mutation point position set; The voltage mutation recognition threshold is set by calculating the mean and standard deviation of the voltage first-order derivative change amplitude sequence within the statistical window and combining it with the sensitivity adjustment coefficient k; S103: Call the timestamps of the temperature derivative sign change position set and the voltage mutation point position set, calculate each pair of time differences, convert the time differences into phase angle values ​​in combination with the current sliding window length, record the window number, and generate a time series window sequence.

4. The insulation strength testing method of an environmentally friendly high-voltage electrical appliance according to claim 1, characterized in that: The specific steps of S2 are: S201: Calculating a phase difference sequence between two adjacent windows based on the time series window sequence, and processing data corresponding to all window numbers to obtain a phase difference change sequence; S202: calling the phase difference change value sequence, calculating the positive sequence statistic value of the window in the sequence by the Mann-Kendall trend test method, and calculating the reverse sequence statistic value at the same time, and determining the direction consistency interval according to the cross relationship between the positive sequence statistic value and the reverse sequence statistic value, and obtaining the trend direction consistent number set; S203: According to the trend direction consistent number set, backtrack to obtain the start and end time segments of the corresponding number window in the original time series window phase difference sequence, combine the boundary numbers of the number interval marked with the phase difference direction consistency information in the time segment, and generate a phase offset interval.

5. The insulation strength testing method of an environmentally friendly high-voltage electrical appliance according to claim 4, characterized in that: The Mann-Kendall trend test method is used to calculate the positive sequence statistics of the window in the sequence , using the formula: ； in, Represents the phase difference change value sequence The phase difference change of each element is Represents the window Elements to The inverse weight factor of the window spacing, Represents the window The element and The time attenuation coefficient of the device thermal inertia of the element, n represents the number of phase difference change values ​​included in the current window, and sgn represents the sign function.

6. The insulation strength testing method of an environmentally friendly high-voltage electrical appliance according to claim 1, characterized in that: The specific steps of S3 are: S301: acquiring the leakage current original signal of the corresponding time period according to the phase offset interval, extracting the current at multiple time points and detecting the positions of local maximum points, marking them as current peak points, and generating a leakage current peak point set; S302: calling the leakage current peak point set, sequentially comparing the current magnitudes of the current peak point with the previous peak point, calculating a trend change identification sequence indicating whether the change direction is rising or falling, and analyzing the occurrence positions of continuous direction reversals in the sequence to generate a reverse sliding trend identification set; S303: extracting the corresponding current peak point number according to the reverse sliding trend identification set, and tracing back the original peak point position and time series number information, marking the time number interval of the peak reversal point in each continuous reverse sliding interval, and generating a local abnormal peak interval.

7. The insulation strength testing method of an environmentally friendly high-voltage electrical appliance according to claim 1, characterized in that: The specific steps of S4 are: S401: Collect resistance strain data at multiple time points on the surface of the insulating medium using a resistance strain gauge, calculate the resistance strain change rate within a sliding window, and extract the response wave peak value corresponding to the window and the relaxation time corresponding to the strain decaying to a stable state after the peak value, to obtain a resistance strain characteristic sequence; S402: Based on the resistance strain characteristic sequence, calculate the local trend component of the relaxation time using the STL algorithm, and determine whether there are synchronous segments in which the peak value shows an upward trend and the relaxation time shows a shortening trend in multiple consecutive windows, to obtain the fatigue trend synchronization interval; S403: Perform cross time period judgment based on the fatigue trend synchronization interval, the phase offset interval, and the local abnormal peak interval, filter a set of time nodes that exist in the time period at the same time, and obtain an insulation strength critical risk node.

8. The insulation strength testing method of an environmentally friendly high-voltage electrical appliance according to claim 7, characterized in that: The local trend component of the relaxation time is calculated by the STL algorithm , using the formula: ； in, represents the ath relaxation duration sampling value of the wth window, represents the arithmetic mean of the relaxation time of the w-th window, represents the bth resistance strain characteristic peak in the wth window, represents the total duration of the w-th time window, The equipment characteristic coefficients determined by calibration experiments, , represents the window time span of the w-th window, Represents the baseline relaxation time constant.

9. The insulation strength testing method of an environmentally friendly high-voltage electrical appliance according to claim 1, characterized in that: The method further comprises: S5: Based on the insulation strength critical risk nodes, the k-nearest neighbor algorithm is used to determine the trend consistency of the overlapping windows, and the node groups with consistent trends and gentle fluctuations in the front and rear neighboring windows are screened to generate an insulation strength confidence critical node set; The insulation strength confidence critical node set includes a node time number, a stable fluctuation trend node group, and a node trend consistency number group.

10. The insulation strength testing method of an environmentally friendly high-voltage electrical appliance according to claim 9, characterized in that: The specific steps of S5 are: S501: Based on the insulation strength critical risk node, obtain the resistance strain peak change trend within the overlapping time window and the change direction of the adjacent window, classify them, calculate the Euclidean distance between the current node and the neighboring nodes using the k-nearest neighbor algorithm, and classify them to obtain a trend consistency matching node set; S502: calling the trend consistency matching node set, calculating the difference between the maximum and minimum values ​​of the resistance strain in the time series corresponding to each group of nodes as the fluctuation amplitude, and filtering the node group by the stability threshold to obtain a stable fluctuation trend node group; The stability threshold is set by the sum of the mean and standard deviation of the resistance strain amplitude in all cross windows; S503: extracting index positions in the time series according to the stable fluctuation trend node group, and performing set mapping processing on the index time points to generate an insulation strength confidence critical node set.

Images