Transient ground potential TEV histogram skewness-kurtosis combined analysis system and algorithm

Abstract

The invention discloses a transient earth potential TEV histogram skewness-kurtosis joint analysis system and algorithm. The system comprises a data acquisition module, a histogram construction module, a feature extraction module, a threshold setting and judgment module, a pulse recognition and positioning module, a result output and suggestion module and an auxiliary function module. The algorithm comprises; the method comprises the steps of 1, data acquisition and preprocessing; step 2, constructing a histogram; step 3, feature extraction; 4, setting and judging a threshold value; step 5, pulse identification and positioning; and step 6, result output and optimization processing. According to the method, the skewness-kurtosis joint analysis algorithm is introduced, the detection capacity of weak discharge signals is improved, the pulse signals are recognized through statistics of signal amplitude distribution characteristics, TEV signals generated by partial discharge can be captured more accurately, and therefore the detection sensitivity is improved, and the detection accuracy is improved. According to the skewness-kurtosis joint analysis algorithm, the influence of electromagnetic interference on an analysis result is reduced by quantifying histogram morphological characteristics.

Description

Technical Field

[0001] This invention relates to the field of charged detection technology, specifically to a system and algorithm for joint analysis of transient ground potential (TEV) histogram skewness and kurtosis. Background Technology

[0002] 10kV-35kV metal-enclosed switchgear is widely used in various substations and ring main units. The safety and reliability of switchgear are crucial to the safe and stable operation of the power distribution network. During long-term operation, metal-enclosed switchgear may experience insulation degradation due to electrical, thermal, and chemical factors, leading to faults. To prevent switchgear failures, real-time monitoring of the switchgear's operating status is necessary. Live-line testing of electrical equipment to detect partial discharge characteristics is currently an effective method for fault prevention. Common live-line testing methods for power distribution switchgear include: 1. Transient ground voltage live-line testing (TEV): capturing high-frequency transient voltages (3-100MHz) induced by partial discharge in the equipment's metal casing; 2. Ultrasonic testing (AE): detecting ultrasonic signals (20-200kHz) generated by partial discharge or mechanical friction; 3. Ultra-high frequency testing (UHF): receiving ultra-high frequency electromagnetic waves (300MHz-3GHz) excited by partial discharge; 4. High-frequency current testing (HFCT): measuring the high-frequency discharge current in the cable grounding wire using a high-frequency current transformer.

[0003] The existing technical solutions have the following drawbacks: First, the detection sensitivity is limited: the TEV method has low sensitivity to certain types of partial discharge (such as surface discharge), and may not be able to accurately detect weak discharge signals. Compared with traditional methods such as pulse current method, its measurement signal-to-noise ratio is low, which may lead to some defects being missed.

[0004] Second, it is highly susceptible to environmental interference: TEV signals are easily affected by external electromagnetic interference (such as frequency converters, wireless communication equipment, etc.), causing fluctuations in histogram data and affecting the accuracy of analysis. During testing, it is necessary to eliminate interference sources or select a suitable testing period.

[0005] Third, insufficient positioning accuracy: TEV testing mainly reflects discharge intensity and frequency, making it difficult to accurately determine the discharge location. For complex equipment structures, it may not be able to accurately locate the specific location of the defect, requiring the use of other testing methods (such as ultrasonic testing) to assist in positioning.

[0006] Fourth, lack of quantitative assessment standards: The quantitative analysis standards for TEV testing are not yet perfect. The test data from different equipment and under different environments lack a unified reference benchmark, which affects the accurate judgment of the severity of defects.

[0007] Fifth, limited ability to identify discharge types: TEV histogram features have limited ability to distinguish different discharge types. Some discharge types (such as floating discharge and tip discharge) may have similar features, which can easily lead to misjudgment. Summary of the Invention

[0008] The purpose of this invention is to provide a system and algorithm for joint analysis of transient ground potential (TEV) histogram skewness and kurtosis, in order to solve the problems mentioned in the background art.

[0009] To achieve the above objectives, the present invention provides the following technical solution: a transient ground potential (TEV) histogram skewness-kurtosis joint analysis system, comprising a data acquisition module, a histogram construction module connected to the data acquisition module, a feature extraction module connected to the histogram construction module, a threshold setting and judgment module connected to the feature extraction module, and a pulse recognition and positioning module connected to the threshold setting and judgment module.

[0010] As a further technical solution of the present invention, the data acquisition module includes a TEV signal acquisition module and a preprocessing module, and the histogram construction module includes an amplitude range module, a grading setting module, a frequency statistics module and a normalization module.

[0011] As a further technical solution of the present invention, the feature extraction module includes a skewness calculation module, a kurtosis calculation module and a tail feature module, and the threshold setting and judgment module includes a three-level threshold setting module, a dynamic baseline adjustment module and a threshold judgment module.

[0012] As a further technical solution of the present invention, the pulse recognition and positioning module includes a pulse detection module and a positioning assistance module.

[0013] As a further technical solution of the present invention, the pulse recognition and positioning module is connected to a result output and suggestion module, which includes a result output module and a suggestion generation module. The result output and suggestion module is connected to an auxiliary function module, which includes a numerical processing optimization module and a smoothing processing module.

[0014] A joint analysis algorithm for skewness-kurtosis of transient ground potential (TEV) histogram includes the following steps: Step 1, data acquisition and preprocessing; Step 2, histogram construction; Step 3, feature extraction; Step 4, threshold setting and judgment; Step 5, pulse identification and localization; and Step 6, result output and optimization.

[0015] In step one above, the data acquisition module uses a high-sensitivity sensor to capture high-frequency transient voltage signals on the metal casing of the switching equipment, and uses a bandpass filter to eliminate external electromagnetic interference and retain the effective signal frequency band.

[0016] In step two above, the signal amplitude characteristics are analyzed by the histogram construction module, a reasonable number of segments is set to balance resolution and computational complexity, the frequency of signals in each amplitude range is counted, and normalization is performed to obtain a probability distribution histogram.

[0017] In step three above, the feature extraction module measures the asymmetry and steepness of the data distribution, analyzes the tail features of the histogram, and identifies significant partial discharge signals.

[0018] In step four above, the threshold setting and judgment module sets three levels of thresholds—normal, suspected abnormal, and abnormal—according to industry standards, and dynamically adjusts the thresholds according to the device type and environment to judge the signal status.

[0019] In step five above, the pulse recognition and positioning module uses a local maximum detection algorithm to identify pulse signals that exceed a threshold, and combines it with other detection methods to improve the accuracy of discharge location positioning.

[0020] In step six above, the results output and suggestion module displays the location, amplitude, and level information of abnormal signals in the form of charts and reports. The auxiliary function module proposes maintenance suggestions based on the detection results and optimizes and smooths the numerical processing.

[0021] As a further technical solution of the present invention, in step three, the skewness of the histogram is calculated by the skewness calculation module in the feature extraction module. Skewness is an indicator of the asymmetry of the probability distribution. Skewness = 0: symmetrical distribution; Skewness > 0: right-skewed distribution with a long tail on the right; Skewness < 0: left-skewed distribution with a long tail on the left. The calculation formula is as follows:

[0022] Skewness = E[(X-μ)] 3 ] / σ 3

[0023] Where μ is the mean, σ is the standard deviation, and E is the expected value operator. The algorithm implementation logic is as follows.

[0024]

[0025]

[0026]

[0027]

[0028] As a further technical solution of the present invention, in step three, the kurtosis in the histogram is extracted by the kurtosis calculation module in the feature extraction module. Kurtosis, as a statistic describing the steepness of the probability distribution, is used to measure the sharpness or flatness of the data distribution. The calculation formula is as follows:

[0029] Kurtosis = [n(n+1)∑(x i -μ) 4 ] / [(n-1)(n-2)(n-3)σ 4 ]-[3(n-1) 2 ] / [(n-2)(n-3)]

[0030] In the formula, n is the number of data points, μ is the mean, σ is the standard deviation, and x i For the i-th data point, the input is a histogram array, where each element represents the count of the corresponding bin. The calculation steps are: calculate the total number of samples n; calculate the mean μ; calculate the second and fourth central moments; apply the kurtosis formula to correct for small samples and output the kurtosis value. Positive kurtosis indicates a peak that is sharper than the normal distribution, negative kurtosis indicates a peak that is flatter than the normal distribution, and zero kurtosis indicates a distribution similar to the normal distribution. The algorithm implementation logic is as follows.

[0031]

[0032]

[0033]

[0034]

[0035] As a further technical solution of the present invention, in step four, the three-level threshold setting module in the threshold setting and judgment module sets the following threshold levels based on the power industry standard DL / T 1630-2016 "Technical Guidelines for Transient Ground Voltage Partial Discharge Detection of Switchgear" and related operating experience: Normal range: background noise level, mean ±2σ; Suspected abnormality: signal of concern, mean +2σ to +5σ; Abnormality: obvious discharge signal, mean +5σ or above; Gaussian distribution 3σ principle expansion is adopted, the mean represents the background noise level, the standard deviation σ reflects the degree of data dispersion, the dynamic baseline adjustment module can adjust the σ coefficient according to the equipment type, and the threshold judgment module compares the calculated skewness and kurtosis values ​​with the set thresholds to determine whether the signal belongs to the normal, suspected abnormal or abnormal range. The algorithm implementation logic is as follows.

[0036]

[0037]

[0038]

[0039]

[0040]

[0041]

[0042] As a further technical solution of the present invention, in step five, a threshold is set by the pulse detection module in the pulse recognition and positioning module to determine whether there is a significant pulse signal. A local maximum detection algorithm is used to identify pulse signals that exceed the threshold. The logical algorithm is as follows:

[0043]

[0044]

[0045]

[0046]

[0047] Furthermore, by combining ultrasonic detection and high-frequency current detection methods with a positioning assistance module, the discharge location is located, thereby improving positioning accuracy and solving the problem of insufficient positioning accuracy in TEV detection.

[0048] Compared with existing technologies, the beneficial effects of this invention are as follows: This invention improves the detection capability of weak discharge signals by introducing a skewness-kurtosis joint analysis algorithm. This algorithm identifies pulse signals by statistically analyzing signal amplitude distribution characteristics, enabling more accurate capture of TEV signals generated by partial discharge, thereby improving detection sensitivity. The skewness-kurtosis joint analysis algorithm reduces the impact of electromagnetic interference on the analysis results by quantifying histogram morphology characteristics. Furthermore, the algorithm effectively reduces the misjudgment rate caused by environmental interference through dual constraints of skewness and kurtosis. Combined with the skewness-kurtosis joint analysis algorithm, the characteristics of discharge signals can be identified more accurately, thus providing a basis for subsequent positioning work. This invention provides more reliable auxiliary information and can be combined with other detection methods to further improve positioning accuracy. Furthermore, it proposes a three-level threshold setting standard based on skewness and kurtosis, providing a quantitative evaluation basis for TEV detection. By setting three levels of thresholds—normal range, suspected abnormality, and abnormality—a unified reference and quantitative evaluation of detection data is achieved. The skewness-kurtosis joint analysis algorithm enhances the algorithm's ability to identify different discharge types by quantifying histogram morphological features. This algorithm can more accurately distinguish the characteristics of different discharge types, reduce the false positive rate, and more accurately identify the characteristics of partial discharge signals, providing strong support for subsequent fault diagnosis and prevention. Attached Figure Description

[0049] Figure 1 This is a system structure diagram of the present invention;

[0050] Figure 2 This is a module architecture diagram of the histogram construction module of the present invention;

[0051] Figure 3 This is a module architecture diagram of the feature extraction module of the present invention;

[0052] Figure 4 This is a module architecture diagram of the threshold setting and judgment module of the present invention;

[0053] Figure 5 This is a system flowchart of the present invention;

[0054] Figure 6 This is a flowchart of the method of the present invention;

[0055] Figure 7 This is an example of histogram analysis in Example 2.

[0056] The diagram shows: 1. Data acquisition module; 11. TEV signal acquisition module; 12. Preprocessing module; 2. Histogram construction module; 21. Amplitude range module; 22. Grading setting module; 23. Frequency statistics module; 24. Normalization module; 3. Feature extraction module; 31. Skewness calculation module; 32. Kurtosis calculation module; 33. Tail feature module; 4. Threshold setting and judgment module; 41. Three-level threshold setting module; 42. Dynamic baseline adjustment module; 43. Threshold judgment module; 5. Pulse recognition and positioning module; 51. Pulse detection module; 52. Positioning assistance module; 6. Result output and suggestion module; 61. Result output module; 62. Suggestion generation module; 7. Auxiliary function module; 71. Numerical processing optimization module; 72. Smoothing module. Detailed Implementation

[0057] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0058] Please see the appendix Figure 1 -Appendix Figure 7 The technical solution provided by this invention:

[0059] Example 1

[0060] A transient ground potential (TEV) histogram skewness-kurtosis joint analysis system includes a data acquisition module 1, a histogram construction module 2 connected to the data acquisition module 1, a feature extraction module 3 connected to the data acquisition module 2, a threshold setting and judgment module 4 connected to the data extraction module 3, and a pulse identification and positioning module 5 connected to the data extraction module 4. The data acquisition module 1 includes a TEV signal acquisition module 11 and a preprocessing module 12. The histogram construction module 2 includes an amplitude range module 21, a grading setting module 22, a frequency statistics module 23, and a normalization module 24. The feature extraction module 3... The system includes a skewness calculation module 31, a kurtosis calculation module 32, and a tail feature module 33. The threshold setting and judgment module 4 includes a three-level threshold setting module 41, a dynamic baseline adjustment module 42, and a threshold judgment module 43. The pulse recognition and positioning module 5 includes a pulse detection module 51 and a positioning assistance module 52. The pulse recognition and positioning module 5 is connected to a result output and suggestion module 6, which includes a result output module 61 and a suggestion generation module 62. The result output and suggestion module 6 is connected to an auxiliary function module 7, which includes a numerical processing optimization module 71 and a smoothing processing module 72.

[0061] Example 2

[0062] Based on a transient ground potential (TEV) histogram skewness-kurtosis joint analysis algorithm, the attached... Figure 7 The analysis was conducted, and the complete analysis process is as follows:

[0063] The image data was converted to linear values, and the antilogarithm of the logarithm base 10 was taken. The results are shown in Table 1. In actual engineering, attention should be paid to the numerical overflow problem, which can be addressed through logarithmic operations or normalization. The mean and standard deviation were calculated based on the linear values: doublemean = 7.69E + 999 (approximately infinite, requiring logarithmic processing in practice); doublestddev = NaN (unable to calculate due to excessively large values); skewness and kurtosis were calculated.

[0064] double skewness=calculateSkewness(histogram);

[0065] skewness = -0.359

[0066] double kurtosis=calculateKurtosis(histogram1)

[0067] kurtosis = 3.469

[0068] Corrected statistical calculations:

[0069] / / Calculate statistics in the logarithmic field

[0070] double[]logYValues={0,0,1,1,2,3,4,3,2,1,0,0,0};

[0071] double[]yMean=10^(mean(logYValues));

[0072] double logMean=calculateMean(yMean); / / ≈833.15

[0073] double logStdDev=calculateStdDev(logYValues,logMean); / / ≈2777.65

[0074] Level 3 threshold (logarithmic domain): Normal threshold: μ + 2σ = 833 + 2 × 2777 ≈ 6388, Abnormal threshold: μ + 5σ = 833 + 5 × 2777 ≈ 14718. The detection results are shown in Table 2. The following conclusions are drawn: The 30dB position (logarithmic value 10000) is judged as "severely abnormal"; the remaining positions are normal. Engineering algorithm correction and numerical processing optimization are needed.

[0075] / / Use logarithmic difference instead of absolute value

[0076] double logThreshold = logMean + 3; / / Equivalent to 1000 times the linear domain

[0077] Example of a dynamic baseline:

[0078]

[0079] Final judgment: If the value at position 30 is greater than 100 times the highest background value → anomaly. In the image, the value at position 30 is 10000, and the background mean is approximately 10 → severe anomaly. Combining this with the skewness value of -0.359 and the kurtosis value of 3.469, the approximate location of the pulse signal distribution in the sample block diagram is slightly to the left, with the distribution in the middle being much sharper than a normal distribution. The data is highly concentrated in a single area, but there are scattered outliers. Considering the linear dB distribution on the horizontal axis, the output result is: position 30 dB (logarithmic value: 10000); judged as a severe anomaly (exceeding 1000 times the background noise). The output recommends immediate ultrasonic testing to check for floating discharge in the switchgear. Through corrected logarithmic domain processing, it can be confirmed that there is a clear abnormal signal at position 30 in this image sample. (Attached) Figure 7 It conforms to typical partial discharge characteristics.

[0080] Table 1 Conversion Results

[0081]

[0082]

[0083] Table 2 Detection Results

[0084]

[0085] Based on the above, the advantages of the present invention are as follows: When using the present invention for the joint analysis of skewness and kurtosis of the transient ground potential (TEV) histogram, the high-frequency transient voltage signal on the metal casing of the switchgear is first captured by the TEV signal acquisition module 11 in the data acquisition module 1, and external electromagnetic interference is eliminated by the preprocessing module 12, retaining the effective signal frequency band; the signal amplitude characteristics are analyzed by the histogram construction module 2, and a reasonable number of segments is set by the amplitude range module 21 and the segment setting module 22 to balance the resolution and computational complexity; the frequency of the signal in each amplitude interval is counted by the frequency statistics module 23, and the probability distribution histogram is obtained by normalization by the normalization module 24; the skewness of the histogram is calculated by the skewness calculation module 31 in the feature extraction module 3, which is an indicator of the asymmetry of the probability distribution. Skewness = 0: symmetrical distribution; skewness > 0: right-skewed distribution with a long tail on the right; skewness < 0: left-skewed distribution with a long tail on the left. The calculation formula is:

[0086] Skewness = E[(X-μ)] 3 ] / σ 3

[0087] Where μ is the mean, σ is the standard deviation, and E is the expected value operator, the kurtosis in the histogram is extracted by the kurtosis calculation module 32 in the feature extraction module 3. Kurtosis, as a statistic describing the steepness of the probability distribution, is used to measure the sharpness or flatness of the data distribution. The calculation formula is:

[0088] Kurtosis = [n(n+1)∑(x i -μ) 4 ] / [(n-1)(n-2)(n-3)σ 4 ]-[3(n-1) 2 ] / [(n-2)(n-3)]

[0089] In the formula, n is the number of data points, μ is the mean, σ is the standard deviation, and x i For the i-th data point, the input is: a histogram array representing the histogram, where each element represents the count of the corresponding bin. The calculation steps are: calculate the total number of samples n; calculate the mean μ; calculate the second and fourth central moments; and apply the kurtosis formula.

[0090] The small sample is corrected, and the kurtosis value is output. Positive kurtosis indicates a peak that is sharper than the normal distribution, negative kurtosis indicates a peak that is flatter than the normal distribution, and zero kurtosis indicates a distribution similar to the normal distribution. The tail feature module 33 analyzes the tail features of the histogram to identify significant partial discharge signals. The three-level threshold setting module 41 in the threshold setting and judgment module 4 is based on the power industry standard DL / T. Based on the "Technical Guidelines for Transient Ground Voltage Partial Discharge Detection of Switchgear" (1630-2016) and relevant operational experience, the following threshold levels are set: Normal range: background noise level, mean ±2σ; Suspected anomaly: signal of concern, mean +2σ to +5σ; Abnormal: obvious discharge signal, mean +5σ or higher. A Gaussian distribution 3σ principle extension is adopted, where the mean represents the background noise level, and the standard deviation σ reflects the data dispersion. The dynamic baseline adjustment module 42 can adjust the σ coefficient according to the equipment type. The threshold judgment module 43 compares the calculated skewness and kurtosis values ​​with the set thresholds to determine whether the signal belongs to the normal, suspected, or abnormal range. The pulse detection module 51 in the pulse recognition and positioning module 5 sets a threshold to determine whether a significant pulse signal exists. A local maximum detection algorithm is used to identify pulse signals exceeding the threshold, and positioning assistance is used to further assist in the determination. The auxiliary module 52 combines ultrasonic detection and high-frequency current detection methods to assist in locating the discharge position, improve positioning accuracy, and solve the problem of insufficient positioning accuracy in TEV detection. The result output module 61 in the result output and suggestion module 6 displays the location, amplitude, and level information of abnormal signals in the form of charts and reports. The suggestion generation module 62 generates corresponding maintenance suggestions based on the detection results. For example, for severe abnormal signals, it is recommended to immediately conduct ultrasonic joint detection to check whether there are problems such as floating discharge in the switch cabinet. The numerical processing optimization module 71 in the auxiliary function module 7 adopts logarithmic domain statistical methods to avoid numerical overflow for large numerical processing and uses logarithmic difference to replace absolute value for calculation to improve calculation stability. The smoothing module 72 smooths the detection data to reduce the impact of random fluctuations on the analysis results and uses algorithms such as moving average and exponential smoothing for data smoothing.

[0091] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.

Claims

1. A transient ground potential (TEV) histogram skewness-kurtosis joint analysis system, comprising a data acquisition module (1), characterized in that: The data acquisition module (1) is connected to a histogram construction module (2), the histogram construction module (2) is connected to a feature extraction module (3), the feature extraction module (3) is connected to a threshold setting and judgment module (4), and the threshold setting and judgment module (4) is connected to a pulse recognition and positioning module (5).

2. The transient ground potential (TEV) histogram skewness-kurtosis joint analysis system according to claim 1, characterized in that: The data acquisition module (1) includes a TEV signal acquisition module (11) and a preprocessing module (12), and the histogram construction module (2) includes an amplitude range module (21), a grading setting module (22), a frequency statistics module (23), and a normalization module (24).

3. The transient ground potential (TEV) histogram skewness-kurtosis joint analysis system according to claim 1, characterized in that: The feature extraction module (3) includes a skewness calculation module (31), a kurtosis calculation module (32), and a tail feature module (33). The threshold setting and judgment module (4) includes a three-level threshold setting module (41), a dynamic baseline adjustment module (42), and a threshold judgment module (43).

4. The transient ground potential (TEV) histogram skewness-kurtosis joint analysis system according to claim 1, characterized in that: The pulse recognition and positioning module (5) includes a pulse detection module (51) and a positioning assistance module (52).

5. The transient ground potential (TEV) histogram skewness-kurtosis joint analysis system according to claim 4, characterized in that: The pulse recognition and positioning module (5) is connected to a result output and suggestion module (6). The result output and suggestion module (6) includes a result output module (61) and a suggestion generation module (62). The result output and suggestion module (6) is connected to an auxiliary function module (7). The auxiliary function module (7) includes a numerical processing optimization module (71) and a smoothing processing module (72).

6. A joint analysis algorithm for transient ground potential (TEV) histogram skewness-kurtosis, comprising: Step 1, data acquisition and preprocessing; Step 2, histogram construction; Step 3, feature extraction; Step 4, threshold setting and judgment; Step 5, pulse identification and localization; Step 6, result output and optimization processing; characterized in that: In step one above, the data acquisition module (1) uses a high-sensitivity sensor to capture the high-frequency transient voltage signal on the metal casing of the switching equipment, and eliminates external electromagnetic interference through a bandpass filter to retain the effective signal frequency band. In step two above, the signal amplitude characteristics are analyzed by the histogram construction module (2), a reasonable number of segments is set to balance the resolution and computational complexity, the frequency of the signal in each amplitude range is counted, and normalization is performed to obtain the probability distribution histogram. In step three above, the asymmetry and steepness of the data distribution are measured by the feature extraction module (3), the tail features of the histogram are analyzed, and significant partial discharge signals are identified. In step four above, the threshold setting and judgment module (4) sets three levels of thresholds: normal, suspected abnormal and abnormal, according to industry standards, and dynamically adjusts the thresholds according to the equipment type and environment to judge the signal status. In step five above, the pulse recognition and positioning module (5) uses a local maximum detection algorithm to identify pulse signals that exceed the threshold, and combines it with other detection methods to improve the accuracy of discharge location positioning. In step six above, the result output and suggestion module (6) displays the location, amplitude, and level information of the abnormal signal in the form of charts and reports, and the auxiliary function module (7) proposes maintenance suggestions based on the detection results and optimizes and smooths the numerical processing.

7. The algorithm for joint analysis of transient ground potential (TEV) histogram skewness and kurtosis according to claim 6, characterized in that: In step three, the skewness of the histogram is calculated by the skewness calculation module (31) in the feature extraction module (3). Skewness is an indicator of the asymmetry of the probability distribution. Skewness = 0: symmetrical distribution; skewness > 0: right-skewed distribution with a long tail on the right; skewness < 0: left-skewed distribution with a long tail on the left. The calculation formula is as follows: Where μ is the mean, σ is the standard deviation, and E is the expected value operator. The algorithm implementation logic is as follows.

8. The algorithm for joint analysis of transient ground potential (TEV) histogram skewness and kurtosis according to claim 7, characterized in that: In step three, the kurtosis in the histogram is extracted by the kurtosis calculation module (32) in the feature extraction module (3). Kurtosis is used as a statistic to describe the steepness of the probability distribution and to measure the sharpness or flatness of the data distribution. The calculation formula is as follows: Kurtosis = [n(n + 1)∑(x i - μ) 4 / [(n - 1)(n - 2)(n - 3)σ 4 -[3(n - 1) 2 / [(n - 2)(n - 3)] In the formula, n is the number of data points, μ is the mean, σ is the standard deviation, and xi is the i-th data point. Input: the histogram array represents the histogram, and each element represents the count of the corresponding bin. Calculation steps: calculate the total number of samples n; calculate the mean μ; calculate the second and fourth central moments; apply the kurtosis formula to correct the small samples and output the kurtosis value. Positive kurtosis indicates a peak that is sharper than the normal distribution, negative kurtosis indicates a peak that is flatter than the normal distribution, and zero kurtosis indicates a distribution similar to the normal distribution. The algorithm implementation logic is as follows.

9. The algorithm for joint analysis of transient ground potential (TEV) histogram skewness and kurtosis according to claim 6, characterized in that: In step four, the three-level threshold setting module (41) in the threshold setting and judgment module (4) sets the following threshold levels based on the power industry standard DL / T 1630-2016 "Technical Guidelines for Transient Ground Voltage Partial Discharge Detection of Switchgear" and related operating experience: Normal range: background noise level, mean ±2σ; Suspected abnormality: signal worth noting, mean +2σ to +5σ; Abnormality: obvious discharge signal, mean +5σ or above; Gaussian distribution 3σ principle expansion is adopted, the mean represents the background noise level, the standard deviation σ reflects the data dispersion, the dynamic baseline adjustment module (42) can adjust the σ coefficient according to the equipment type, and the threshold judgment module (43) compares the calculated skewness and kurtosis values ​​with the set thresholds to determine whether the signal belongs to the normal, suspected abnormal or abnormal range. The algorithm implementation logic is as follows.

10. The algorithm for joint analysis of transient ground potential (TEV) histogram skewness and kurtosis according to claim 6, characterized in that: In step five, a threshold is set by the pulse detection module (51) in the pulse recognition and positioning module (5) to determine whether there is a significant pulse signal. The local maximum detection algorithm is used to identify pulse signals that exceed the threshold. The logical algorithm is as follows: Furthermore, by combining ultrasonic detection and high-frequency current detection methods with the positioning assistance module (52), the discharge position is located, improving the positioning accuracy and solving the problem of insufficient positioning accuracy of TEV detection.

Images