A method for calculating acoustic-magnetic time difference based on signal quality adaptation

Abstract

The application provides a kind of acoustic-magnetic time difference calculation method based on signal quality self-adaption, belongs to cable fault detection technical field.The method includes: obtaining the acoustic signal and electromagnetic signal of high-voltage cable partial discharge, detecting electromagnetic signal to determine trigger time;With the time as a reference, the sound signal segment and background noise signal segment are intercepted, and the signal-to-noise ratio and maximum peak-to-peak value are calculated after preprocessing;Compare it with the preset threshold, dynamically select the trigger point positioning scheme;Determine the effective trigger point according to the scheme, calculate the original acoustic-magnetic time difference;Calculate the energy of the sound signal this time, and correct the original time difference combined with the reference energy to get the final time difference.The method improves the accuracy and reliability of time difference calculation by adaptive matching scheme, adapts to different signal quality scenarios, and enhances the application adaptability of complex field environment.

Description

Technical Field

[0001] This invention relates to the field of cable fault detection technology, and more specifically to a method for calculating the acousto-magnetic time difference based on signal quality adaptation. Background Technology

[0002] High-voltage cables are a critical component of power transmission networks, and their operational reliability directly impacts the safe supply of electricity to the grid. Partial discharge is a significant indicator of cable insulation degradation and a major cause of faults; therefore, accurate detection and location of partial discharges are a core requirement for ensuring the safety of power equipment. Acoustomagnetic time-difference positioning (ADTMS) technology has become the mainstream solution for locating partial discharges in high-voltage cables due to its high accuracy and wide applicability. Its core principle is to capture the time difference between the arrival of the acoustic and electromagnetic signals generated by the partial discharge, and then use the difference in their propagation speeds to deduce the location of the fault.

[0003] Existing acoustomagnetic positioning technology has significant shortcomings in practical applications: prominent electromagnetic interference in the field, complex background noise with large fluctuations in complex industrial, underground or underwater scenarios, resulting in low signal-to-noise ratio of acoustic signals, inaccurate trigger point judgment, and reduced positioning reliability; at the same time, existing solutions mostly focus on hardware filtering optimization or post-processing of data, lacking the ability to adapt to dynamic changes in signal quality, and have not established a hierarchical processing mechanism for different signal-to-noise ratio scenarios, making it difficult to flexibly adapt to diverse application environments and solve the problem of accurate positioning under complex noise. Summary of the Invention

[0004] This invention provides an acoustomagnetic time difference calculation method based on signal quality adaptation, which solves the problems in the prior art such as low signal-to-noise ratio of acoustic signal leading to trigger point judgment deviation, insufficient positioning accuracy and reliability, and lack of adaptive capability and hierarchical processing mechanism for dynamic changes in signal quality.

[0005] To achieve the above objectives, this invention provides a method for calculating the acoustic-magnetic time difference (AMTD) based on adaptive signal quality, comprising the following steps: Signal acquisition and trigger benchmark establishment step: acquiring acoustic and electromagnetic signals generated by partial discharge of a high-voltage cable, and performing trigger detection on the electromagnetic signals to determine the trigger time; Signal preprocessing and quality assessment step: using the trigger time of the electromagnetic signals as a benchmark, extracting a sound signal segment and a background noise signal segment of a preset time length; preprocessing the sound signal segment and the background noise signal segment, and calculating the signal-to-noise ratio (SNR) and maximum peak-to-peak value (MPV) based on the preprocessed signals; Adaptive scheme selection step: comparing the calculated SNR and MPV with preset threshold conditions, and dynamically selecting the corresponding trigger point positioning scheme based on the comparison results; Trigger point positioning step: determining the effective trigger point of the sound signal according to the selected positioning scheme; Time difference calculation and correction step: calculating the energy of the acquired acoustic signal based on the sound signal segment; calculating the original AMTD based on the effective trigger point and the trigger point of the electromagnetic signals, and performing energy adaptive correction on the original AMTD by combining the sound signal energy with a preset reference energy to obtain the final AMTD.

[0006] Optionally, the trigger detection includes: detecting whether the amplitude of the electromagnetic signal exceeds a preset fixed threshold, and determining the moment when the amplitude first exceeds the fixed threshold as the trigger moment of the electromagnetic signal.

[0007] Optionally, the preset threshold conditions include a first signal-to-noise ratio (SNR) threshold, a second SNR threshold, a first peak-to-peak value (PVP) threshold, and a second PVP threshold; the trigger point localization scheme includes a direct peak localization scheme, a similar extreme value matching localization scheme, and a cumulative average localization scheme. The step of dynamically selecting the corresponding trigger point localization scheme based on the comparison results includes: if the SNR is greater than the first SNR threshold and the maximum PVP is greater than the first PVP threshold, then the direct peak localization scheme is selected; if the SNR is not greater than the first SNR threshold but is greater than the second SNR threshold, and the maximum PVP is greater than the second PVP threshold, then the similar extreme value matching localization scheme is selected; otherwise, the segmented correlation and cumulative average localization scheme is selected.

[0008] Optionally, the direct peak positioning scheme includes: in the sound signal segment, determining the first peak point whose amplitude exceeds a preset multiple of the maximum peak value as the effective trigger point.

[0009] Optionally, the similar extreme value matching and localization scheme includes: an extreme value matching step: performing distribution matching between the extreme points of the acquired sound signal and the reference sound signal; wherein, the distribution matching includes: matching the minimum points of the two and retaining the minimum point pairs with positional deviations within a preset range, locating the maximum points between adjacent minimum points based on the minimum point pairs, constructing their respective extreme value sequences, calculating the positional difference between the corresponding maximum point and the adjacent minimum point in the two sets of sequences, and retaining the maximum points with positional difference deviations within a preset range as the matched extreme value point pairs; a gradient judgment step: calculating the bilateral gradient symmetry score for the candidate peak points in the matched extreme value point pairs; and a trigger point judgment step: determining the first candidate peak point with a bilateral gradient symmetry score exceeding a preset score threshold as the effective trigger point.

[0010] Optionally, the calculation process of the two-sided gradient symmetry score includes: taking multiple sampling points before and after the candidate peak point, calculating the sign consistency ratio of the forward gradient and the backward gradient, and using the ratio as the two-sided gradient symmetry score.

[0011] Optionally, the segmented correlation and cumulative averaging localization scheme includes: a signal enhancement step: performing spectral subtraction enhancement processing on the sound signal segment based on the background noise signal segment; a segmented filtering step: performing segmented cross-correlation analysis on the enhanced multiple sound signals, calculating the segmented cross-correlation coefficients between the multiple sound signals, and removing abnormal signal segments with minimum correlation coefficients lower than the median and differences exceeding a preset difference to obtain filtered signal segments; and a cumulative averaging step: performing cumulative averaging processing on the filtered signal segments, and determining the effective trigger point from the cumulatively averaged signal.

[0012] Optionally, the spectral enhancement process includes: dividing the background noise signal segment into segments with a preset frame length, selecting stable noise frames, estimating the noise power spectrum based on the noise frames, and subtracting a fixed multiple of the noise power spectrum from the power spectrum of the sound signal segment to achieve signal enhancement.

[0013] Optionally, the step of calculating the current acoustic signal energy based on the acoustic signal segment specifically involves: calculating the sum of squares of the amplitudes of all sampling points in the acoustic signal segment, and using the sum of squares as the current acoustic signal energy.

[0014] Optionally, the reference energy is the acoustic signal energy corresponding to the smallest acousto-magnetic time difference in historical detection data.

[0015] The present invention provides an adaptive acoustic-magnetic time difference (AMTD) calculation method based on signal quality. By preprocessing and quality assessment of the acoustic signal and background noise, it dynamically matches hierarchical localization schemes for different signal-to-noise ratio scenarios. Combined with signal enhancement strategies such as extreme value matching and segmented correlation cumulative averaging, and an energy adaptive correction mechanism, it effectively suppresses noise interference in complex environments, avoids misjudgment of trigger points, and significantly improves the accuracy and stability of AMTD calculation. At the same time, it has strong environmental adaptability and can flexibly cope with diverse application scenarios from ideal laboratories to strong interference sites, providing reliable technical support for the accurate localization of partial discharge faults in high-voltage cables. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. In the drawings:

[0017] Figure 1 This is a flowchart of the acoustic-magnetic time difference calculation method provided in the embodiments of the present invention;

[0018] Figure 2 This is a logic diagram for selecting the trigger point location scheme provided in the embodiments of the present invention;

[0019] Figure 3 This is a flowchart of the trigger point location scheme selection provided in the embodiments of the present invention;

[0020] Figure 4 This is a flowchart of the similar extreme value matching and localization scheme provided in the embodiments of the present invention. Detailed Implementation

[0021] The specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are for illustration and explanation only and are not intended to limit the scope of the present invention.

[0022] It should be noted that the acquisition, transmission, storage, use, and processing of data in the technical solution of this application all comply with the relevant provisions of national laws and regulations. In the embodiments of this application, certain existing industry solutions such as software, components, and models may be mentioned. These should be considered exemplary, intended only to illustrate the feasibility of implementing the technical solution of this application, and do not imply that the applicant has already used or necessarily used such solutions.

[0023] With the increasing demands for reliability and fault location accuracy in high-voltage cables from power systems, existing acousto-magnetic time difference (TDOT) localization technologies suffer from problems such as low signal-to-noise ratio due to complex environmental noise, trigger point judgment errors, lack of dynamic adaptive signal quality capabilities, and absence of differentiated hierarchical processing mechanisms. Therefore, developing an adaptive processing scheme that adapts to various noise scenarios, suppresses interference, and improves the accuracy of time difference calculation is crucial.

[0024] To address this issue, this invention proposes an adaptive acoustic-magnetic time difference (AMTD) calculation method. This method constructs a quantitative basis through preprocessing and quality assessment of the acoustic signal and background noise. It dynamically selects a graded positioning scheme based on the signal-to-noise ratio (SNR) and peak-to-peak value, combining extreme value matching, segmented correlation averaging, and other signal enhancement strategies with an energy adaptive correction mechanism. This forms a closed-loop process from signal acquisition, quality assessment, scheme adaptation to time difference correction, significantly improving the accuracy, environmental adaptability, and positioning reliability of AMTD calculation, providing technical support for accurate location of partial discharge faults in high-voltage cables.

[0025] The following is combined Figures 1-4 This invention is described in detail.

[0026] like Figure 1 As shown, this embodiment of the invention provides a method for calculating the acoustic-magnetic time difference based on signal quality adaptation, including the following steps: Signal acquisition and trigger benchmark establishment step: Acquire acoustic and electromagnetic signals generated by partial discharge of high-voltage cables, and perform trigger detection on the electromagnetic signals to determine the trigger time of the electromagnetic signals; Signal preprocessing and quality assessment step: Based on the trigger time of the electromagnetic signals, extract a sound signal segment and a background noise signal segment of a preset time length; Preprocess the sound signal segment and the background noise signal segment, and calculate the signal-to-noise ratio and maximum peak-to-peak value based on the preprocessed signals; Adaptive scheme selection step: Compare the calculated signal-to-noise ratio and maximum peak-to-peak value with preset threshold conditions, and dynamically select the corresponding trigger point positioning scheme according to the comparison results; Trigger point positioning step: Determine the effective trigger point of the sound signal according to the selected positioning scheme; Time difference calculation and correction step: Calculate the energy of the acoustic signal acquired this time based on the sound signal segment; Calculate the original acoustic-magnetic time difference based on the effective trigger point and the trigger point of the electromagnetic signals, and combine the sound signal energy with the preset reference energy to perform energy adaptive correction on the original acoustic-magnetic time difference to obtain the final acoustic-magnetic time difference.

[0027] Among them, the acoustic signal is the ultrasonic signal caused by the partial discharge of high-voltage cables, and the electromagnetic signal is the electromagnetic pulse signal generated by the discharge. The synchronous acquisition of the two types of signals is the basis for time difference calculation. The sound signal segment is the segment containing the target acoustic signal after the triggering time of the electromagnetic signal, and the background noise signal segment is the segment containing only environmental interference before the triggering time. Preprocessing is used to improve signal quality. The signal-to-noise ratio quantifies the intensity of the acoustic signal relative to the noise. The maximum peak-to-peak value reflects the range of signal amplitude fluctuation. The preset threshold conditions include two sets of signal-to-noise ratio thresholds and two sets of peak-to-peak value thresholds, corresponding to three types of signal quality levels. The core of dynamically selecting the positioning scheme is to balance detection efficiency and accuracy. The effective trigger point is a characteristic point that reflects the initial propagation time of the acoustic signal. Its accuracy directly determines the accuracy of time difference calculation. The acoustic signal energy reflects the signal strength and propagation attenuation. The reference energy is the acoustic signal energy corresponding to the minimum time difference in historical data. Energy adaptive correction is used to compensate for calculation deviations caused by changes in propagation path and medium characteristics.

[0028] Specifically, acoustic and electromagnetic signals are captured by adapted sensors. Sensor deployment must ensure effective reception of the target signal. The amplitude of the electromagnetic signal is continuously monitored, and the moment when the amplitude first exceeds a preset fixed threshold is recorded as the trigger moment of the electromagnetic signal. Preprocessing includes low-pass filtering, downsampling, removal of DC components, and elimination of noise extrema based on the statistical distribution of background noise. The maximum peak-to-peak value is the normalized relative intensity index. When the signal quality is high, an efficient positioning scheme is selected; when the quality is medium, a scheme that balances accuracy and anti-interference is selected; and when the quality is low, a strong noise suppression scheme is selected. Based on the selected positioning scheme, effective trigger points that meet the feature requirements are selected from the preprocessed sound signal segments. The acoustic signal energy is obtained by calculating the sum of the squares of the amplitudes of all sampling points in the sound signal segment. The original time difference is the difference between the timestamp of the effective trigger point and the trigger moment of the electromagnetic signal. The correction coefficient is determined by comparing the current energy with the reference energy, and the original time difference is corrected to obtain a high-precision final result.

[0029] Specifically, the signal-to-noise ratio (SNR) calculation formula is as follows:

[0030]

[0031] in, Represents the total power of the audio signal segment. This represents the power of the background noise signal segment.

[0032] The acoustic-magnetic time difference (AMTD) calculation method based on signal quality adaptation provided in this invention synchronously captures the acoustic and electromagnetic signals generated by partial discharge in high-voltage cables and establishes a unified triggering benchmark. After preprocessing and quality assessment, the signal-to-noise ratio and maximum peak-to-peak value characteristics are quantified. Based on a trigger point location scheme dynamically matched to the signal quality, and combined with the acoustic signal energy and historical reference energy, the original AMTD is adaptively corrected. This effectively suppresses noise interference in complex environments, solves the problem of trigger point judgment deviation caused by dynamic fluctuations in signal quality, and significantly improves the accuracy and stability of AMTD calculation. At the same time, it has the ability to flexibly adapt to diverse application scenarios from ideal environments to strong interference sites, providing reliable technical support for the accurate location of partial discharge faults in high-voltage cables.

[0033] Preferably, the trigger detection includes: detecting whether the amplitude of the electromagnetic signal exceeds a preset fixed threshold, and determining the moment when the amplitude first exceeds the fixed threshold as the trigger moment of the electromagnetic signal.

[0034] Among them, the fixed threshold is a preset judgment threshold based on the normal fluctuation range of electromagnetic signals and the characteristic amplitude of partial discharge electromagnetic pulse signals. Its core function is to distinguish between small-amplitude signals generated by environmental interference and effective electromagnetic signals caused by partial discharge, ensuring the accuracy of triggering time determination. The core objective of trigger detection is to accurately capture the starting propagation time of electromagnetic signals, providing a unified and reliable time reference for subsequent interception of acoustic signal segments and background noise signal segments and time difference calculation.

[0035] Specifically, the system continuously monitors the amplitude of the captured electromagnetic signals in real time. The setting of the fixed threshold needs to be calibrated in combination with the on-site electromagnetic environment, cable voltage level, and typical amplitude characteristics of partial discharge signals to ensure that valid signals are not missed due to excessively high thresholds, nor are interference signals misjudged as valid signals due to excessively low thresholds. When the amplitude of the electromagnetic signal is detected to rise from below the threshold for the first time and exceed the fixed threshold, the moment is immediately recorded and determined as the electromagnetic signal trigger moment, thus completing the establishment of the trigger reference.

[0036] The preferred embodiment of the present invention provides a technical solution that accurately locks the trigger time of the electromagnetic signal, avoids misjudgment and missed detection caused by interference, and provides a stable and reliable time reference for subsequent signal processing and time difference calculation.

[0037] like Figure 2 and Figure 3As shown, preferably, the preset threshold conditions include a first signal-to-noise ratio (SNR) threshold, a second SNR threshold, a first peak-to-peak value (PFV) threshold, and a second PFV threshold; the trigger point localization scheme includes a direct peak localization scheme, a similar extreme value matching localization scheme, and a cumulative average localization scheme. The step of dynamically selecting the corresponding trigger point localization scheme based on the comparison results includes: if the SNR is greater than the first SNR threshold and the maximum PFV value is greater than the first PFV threshold, then the direct peak localization scheme is selected; if the SNR is not greater than the first SNR threshold but is greater than the second SNR threshold, and the maximum PFV value is greater than the second PFV threshold, then the similar extreme value matching localization scheme is selected; otherwise, the segmented correlation and cumulative average localization scheme is selected.

[0038] Among them, the first signal-to-noise ratio threshold, the second signal-to-noise ratio threshold, the first peak-to-peak threshold, and the second peak-to-peak threshold constitute the signal quality grading standard, which is used to distinguish between high, medium, and low signal quality. The three positioning schemes are adapted to signals with different levels of noise interference. The core is to balance detection efficiency and anti-interference capability through dynamic matching.

[0039] Specifically, the first signal-to-noise ratio (SNR) threshold is higher than the second SNR threshold, and the first peak-to-peak value (PVP) threshold is higher than the second PVP threshold. When a signal simultaneously meets the conditions of having an SNR exceeding the first SNR threshold and a PVP exceeding the first PVP threshold, it is determined to be a high-quality signal and the direct peak localization scheme is selected. If the above conditions are not met, but the signal-to-noise ratio exceeds the second SNR threshold and the PVP exceeds the second PVP threshold, it is determined to be a medium-quality signal and the similar extreme value matching localization scheme is selected. In other cases, it is a low-quality signal, and the segmented correlation and cumulative averaging localization scheme is selected to adapt to the signal processing needs of different scenarios.

[0040] Specifically, the signal peak value is all the extreme points extracted from the preprocessed sig_voice. It should be noted that because this process uses MATLAB's audioread function to read the audio file without using the 'native' option, the read signal data has been automatically normalized to the [-1, 1] interval. Based on this, the maximum peak-to-peak value is a dimensionless relative intensity index, theoretically not exceeding 2.

[0041] Taking the on-site detection of partial discharge in a high-voltage cable as an example, the first signal-to-noise ratio (SNR) threshold is preset to 40dB, the second SNR threshold to 20dB, the first peak-to-peak value threshold to 0.1, and the second peak-to-peak value threshold to 0.01. When the detected acoustic signal SNR is 65dB and the maximum peak-to-peak value is 0.2, it is determined to be a high-quality signal and the direct peak localization scheme is selected. When the SNR is 38dB and the maximum peak-to-peak value is 0.08, it is determined to be a medium-quality signal and the similar extreme value matching localization scheme is selected. When the SNR is 15dB and the maximum peak-to-peak value is 0.008, it is determined to be a low-quality signal and the segmented correlation and cumulative averaging localization scheme is selected, thus achieving accurate scheme adaptation under different noise scenarios.

[0042] The preferred embodiment of this invention achieves accurate signal quality determination by setting graded thresholds and dynamically matches three positioning schemes: direct peak positioning, similar extreme value matching, and segmented correlation and cumulative averaging. This ensures detection efficiency in high-quality signal scenarios and enhances anti-interference capabilities in medium- and low-quality signal scenarios, effectively avoiding the limitations of a single scheme in complex environments. It significantly improves the accuracy and reliability of trigger point positioning under different noise interference scenarios, providing solid support for the accurate calculation of subsequent acoustic-magnetic time difference.

[0043] Preferably, the direct peak positioning scheme includes: in the sound signal segment, determining the first peak point whose amplitude exceeds a preset multiple of the maximum peak value as the effective trigger point.

[0044] The preset multiplier is a proportional threshold set based on the amplitude characteristics of the sound signal segment and the noise suppression requirements. Its core function is to filter out effective signal peaks with significant characteristics and eliminate the interference of small-amplitude noise peaks. The peak point refers to the extreme point in the sound signal segment where the amplitude is higher than the surrounding signals. The core advantage of this scheme is its high detection efficiency and adaptability to high-quality signal scenarios.

[0045] Specifically, the preset multiple needs to be calibrated in conjunction with the on-site noise level and the typical amplitude characteristics of the partial discharge sound signal, and is usually set to 0.2. The amplitude of the pre-processed sound signal segment is scanned, all peak points that meet the local extremum conditions are extracted, and the first peak point whose amplitude exceeds the preset multiple of the maximum peak value is selected in chronological order and directly determined as the effective trigger point, thus quickly completing the trigger point location.

[0046] The direct peak localization scheme provided by the preferred embodiment of the present invention is adapted to high-quality signal scenarios. It filters significant characteristic peak points with a preset multiple, quickly eliminates small-amplitude noise interference, and greatly improves detection efficiency while ensuring the accuracy of trigger point localization, providing efficient support for rapid calculation of acoustic-magnetic time difference.

[0047] like Figure 4As shown, preferably, the similar extreme value matching and positioning scheme includes: an extreme value matching step: performing distribution matching between the extreme points of the acquired sound signal and the reference sound signal; wherein, the distribution matching includes: matching the minimum points of the two and retaining the minimum point pairs with positional deviations within a preset range, locating the maximum points between adjacent minimum points based on the minimum point pairs, constructing their respective extreme value sequences, calculating the positional difference between the corresponding maximum points and adjacent minimum points in the two sets of sequences, and retaining the maximum points with positional difference deviations within a preset range as the matched extreme value point pairs; a gradient judgment step: calculating the bilateral gradient symmetry score for the candidate peak points in the matched extreme value point pairs; a trigger point judgment step: determining the first candidate peak point with a bilateral gradient symmetry score exceeding a preset score threshold as the effective trigger point.

[0048] More preferably, the calculation process of the two-sided gradient symmetry score includes: taking multiple sampling points before and after the candidate peak point, calculating the sign consistency ratio of the forward gradient and the backward gradient, and using the ratio as the two-sided gradient symmetry score.

[0049] The reference sound signal is a standard effective sound signal of partial discharge from a high-voltage cable of the same type. After noise reduction and validity verification, it serves as a feature matching benchmark to provide a stable signal feature template. The preset range refers to the position deviation threshold calibrated based on the signal sampling rate, the propagation characteristics of the partial discharge sound signal, and the on-site environment. It includes the position deviation range of the minimum point and the position difference deviation range of the maximum point. The core is to ensure the rationality of feature matching. The extreme value sequence is a structured feature set formed by arranging the matched minimum points and the maximum points between adjacent minimum points in chronological order. It is used to establish a one-to-one correspondence between the current signal and the reference signal. The bilateral gradient symmetry score verifies the signal authenticity of the peak point by quantifying the consistency of the rising and falling edge features of the candidate peak point. The entire solution is suitable for medium-quality signal scenarios. The core is to strengthen the suppression capability of medium-intensity noise through the dual mechanism of feature matching and gradient verification.

[0050] Specifically, in the extreme value matching step, a pre-stored reference sound signal is first retrieved. A sliding window method is used to extract all minimum points between the current sound signal and the reference sound signal. The time position deviation of the corresponding minimum points is calculated. The window length is set to 20-30 sampling points according to the sampling rate. Pairs of minimum points with deviations within a preset range are retained. This preset range is typically set to 5-8 sampling points, corresponding to a time deviation of 0.3125-0.5ms. Using these minimum point pairs as anchor points, all maximum points between each pair of minimum points are located. Through spectral analysis of the acquired sound signal, it is found that its effective frequency components are mainly distributed in the 100-300Hz range. Based on this, the key parameters for peak matching are determined: calculated according to the highest frequency component of 300Hz, the number of sampling points corresponding to a single complete oscillation cycle is... ≈53. Considering that in actual waveforms, the transition from a trough to the adjacent peaks typically occupies about two-thirds of the oscillation period, the maximum number of points required to reach the peak on both sides of the trough can be calculated as follows: ≈35, meaning a maximum of 17 points on one side, so no more than 17 points are selected (set to 15). The located maxima points are arranged in chronological order to form extreme value sequences for the current and reference acoustic signals. Each element in the sequence contains the amplitude of the maxima point and the positional difference between it and the preceding and following minima. The positional difference between the corresponding maxima point and the adjacent minima in the two extreme value sequences is calculated. Maxima points with difference deviations within a preset range are retained as matching extreme value pairs to filter out candidate peak points. In the gradient judgment step, 15 sampling points are taken before and after the candidate peak point according to the signal sampling frequency. The forward and backward gradients are calculated, and the number of sampling point pairs with the same sign in the two gradient sets is counted. The ratio of this number to the total number of sampling point pairs is the bilateral gradient symmetry score. In the trigger point judgment step, a preset scoring threshold of 0.7 is used. Candidate peak points are traversed in chronological order, and the first point with a bilateral gradient symmetry score exceeding this threshold is selected as a valid trigger point, achieving accurate positioning under moderate noise interference.

[0051] Minimum point The matching formula is:

[0052]

[0053] The formula for locating the maximum point is:

[0054]

[0055]

[0056] Formula for judging the deviation of the position difference of the maximum point:

[0057]

[0058] Two-sided gradient symmetry scoring formula:

[0059]

[0060] Formula for effective peak value selection criteria:

[0061]

[0062] Where δ is the position matching threshold, which can be customized to allow a deviation of n milliseconds, i.e. , This is the set of locations of the maxima of the reference signal; This is the set of locations of the minimum points of the reference signal; This is the set of locations of the maximum values ​​of the current signal; This is the set of locations of the minimum points of the current signal; The location of a single minimum point in the reference signal; This represents the location of a single minimum point in the current signal. The location of a single maximum point between adjacent minimum points in the reference signal; This represents the location of a single maximum point between adjacent minimum points in the current signal; This represents the location of the k-th minimum point in the reference signal; This represents the location of the (k+1)th local minimum point in the reference signal. This represents the location of the k-th minimum point in the current signal; This represents the position of the (k+1)th local minimum point in the current signal. denoted as the positional difference deviation between the k-th pair of maxima and their respective adjacent minima; Scoring the two-sided gradient symmetry of the candidate peak point p; The number of points for gradient calculation on both sides of the peak value is fixed at 15, and is determined based on the effective frequency of the signal and the sampling rate. The parentheses () are sign functions that return -1, 0, or 1 after a numerical value is entered. () represents the acoustic signal to be analyzed after low-pass filtering, downsampling, and DC component removal preprocessing. The result of the validity determination of the candidate peak point p. The signal sampling rate is fixed at 16kHz to balance signal resolution and computational complexity. The highest effective frequency component of the sound signal is fixed at 300Hz, and the effective frequency range of the signal is determined by spectrum analysis to be 100-300Hz.

[0063] Taking the on-site detection of partial discharge in high-voltage cables as an example, the signal sampling rate is set to 16 kHz. Spectrum analysis determines the highest effective frequency of the sound signal to be 300 Hz. The set of maximum points of the reference signal is {200, 450, 720}, the set of minimum points of the reference signal is {100, 320, 600}, the set of maximum points of the current signal is {202, 453, 722}, and the set of minimum points of the current signal is {102, 321, 601}. The custom allowable time deviation is 0.2 milliseconds. The position matching threshold is calculated as 3.2 based on the formula "position matching threshold equals 16 multiplied by the custom allowable deviation in milliseconds". Three sampling points are selected. First, minimum point matching is performed, filtering out point pairs where the position deviation between the current signal minimum point and the reference signal minimum point is less than 3 sampling points, i.e., {(100,102), (320,321), (600,601)}. Then, based on parameter 15, the maximum point between adjacent minimum points is located. The maximum point in the reference signal that satisfies the condition "between the position of the first minimum point plus 15 sampling points and the position of the second minimum point" is 200. The maximum point in the current signal that satisfies the condition "between the position of the first minimum point plus 15 sampling points and the position of the second minimum point" is 202. Then, the position difference deviation is calculated, which is "the difference between the absolute values ​​of the position difference between the first maximum point and the corresponding minimum point in the reference signal and the position difference between the first maximum point and the corresponding minimum point in the current signal". The result is 0, which is less than the position matching threshold of 3. The extreme value match is determined, and the maximum point 202 of the current signal is taken as a candidate peak point. 15 sampling points are taken before and after the candidate peak point to calculate the forward gradient and backward gradient. The number of sampling point pairs with the same sign in the two sets of gradients is counted. According to the two-sided gradient symmetry scoring rule, the score is 0.85, which meets the effective peak screening condition of "score greater than 0.7". Finally, the point is determined as an effective trigger point, realizing accurate positioning under moderate noise interference.

[0064] The preferred embodiment of this invention provides a technical solution suitable for medium-quality signal scenarios with a signal-to-noise ratio greater than 20dB and a maximum peak-to-peak value greater than 0.01. Through a dual mechanism of "similar extreme value matching + two-sided gradient verification," relying on the effective frequency range of 100-300Hz determined by spectrum analysis and a sampling rate of 16kHz, the peak matching parameter 15 and the number of gradient calculation points 15 are scientifically derived. First, a stable extreme value sequence correspondence is constructed through minimum point matching, maximum point location, and position difference deviation judgment. Then, the peak authenticity is verified through two-sided gradient symmetry scoring. It effectively utilizes the difference between the coherence of the signal extreme value sequence and the randomness of noise, strongly suppresses medium-intensity noise interference, accurately eliminates false peaks, and ensures that in scenarios where the signal is interfered with but the waveform structure is relatively complete, the real and effective trigger points can be quickly screened out. This provides reliable support for the accurate calculation of subsequent acoustic-magnetic time difference and significantly improves the stability and accuracy of partial discharge location in high-voltage cables.

[0065] Preferably, the segmented correlation and cumulative averaging localization scheme includes: a signal enhancement step: performing spectral subtraction enhancement processing on the sound signal segment based on the background noise signal segment; a segmented filtering step: performing segmented cross-correlation analysis on the enhanced multiple sound signals, calculating the segmented cross-correlation coefficients between the multiple sound signals, and removing abnormal signal segments with minimum correlation coefficients lower than the median and differences exceeding a preset difference to obtain the filtered signal segments; and a cumulative averaging step: performing cumulative averaging processing on the filtered signal segments, and determining the effective trigger point from the cumulatively averaged signal.

[0066] More preferably, the spectral enhancement processing includes: dividing the background noise signal segment into segments of a preset frame length, selecting stable noise frames, estimating the noise power spectrum based on the noise frames, and subtracting a fixed multiple of the noise power spectrum from the power spectrum of the sound signal segment to achieve signal enhancement.

[0067] Once the segmented correlation and cumulative averaging positioning scheme is activated, it will be locked in operation under this scheme until the preset exit conditions are met. This scheme is suitable for low-quality signal scenarios that do not meet the conditions for high-quality or medium-quality signals. The core of the scheme is a progressive combination mechanism of spectral subtraction signal enhancement, segmented cross-correlation screening, cumulative averaging noise reduction, and energy coefficient correction. It first solves the problems of insufficient signal purity and low data validity under low signal-to-noise ratio, and then compensates for the deviation caused by the difference in propagation path and medium through energy adaptive correction. At the same time, the system will lock in operation after entering this mode until a valid silencing trigger signal is detected, avoiding frequent scheme switching that may affect stability and ensuring the robustness and accuracy of positioning in low-quality signal scenarios.

[0068] Specifically, the first step is spectral subtraction signal enhancement: the signal 300ms before magnetic field triggering is selected as the noise estimation segment, divided into 100ms frame lengths, and the standard deviation of each frame is calculated. Frames whose difference from the minimum standard deviation is less than 0.1 times the overall median are identified as noise frames. The average power spectrum of the noise frames and the power spectrum of the target sound signal are calculated separately. By subtracting 10 times the noise power spectrum from the signal power spectrum, signal reconstruction and enhancement are achieved, effectively suppressing background noise. The second step is segmented cross-correlation analysis: the enhanced signal is divided into several segments, and the segmented cross-correlation coefficients between multiple acquired signals are calculated for each segment. The minimum value of the correlation coefficient in each segment is extracted. When the difference between this minimum value and the median exceeds 10% of the median, it is identified as an abnormal data packet and discarded to eliminate invalid data caused by strong interference. The third step is cumulative averaging noise reduction: before the mute triggering, the target sound signals acquired multiple times are continuously accumulated and averaged to iteratively improve the signal-to-noise ratio and gradually optimize the signal waveform integrity. The fourth step is energy coefficient correction: the signal energy corresponding to the minimum acoustomagnetic time difference is recorded as the reference energy, the energy of the current target acoustic signal is calculated, and the correction coefficient is calculated based on the order of magnitude difference between the reference energy and the current energy. This coefficient is used to correct the current acoustomagnetic time difference to compensate for the deviation caused by propagation path attenuation. Finally, significant peaks are identified from the accumulated average signal, and combined with the corrected time difference, they are determined as valid trigger points to ensure the consistency and reliability of positioning in low-quality signal scenarios.

[0069] Specifically, the formula for signal power spectrum enhancement is:

[0070]

[0071] The formula for calculating signal energy is:

[0072]

[0073] The formula for calculating the energy correction factor is:

[0074]

[0075] in, For the target sound signal at frequency The power spectrum at that location; For noisy frames at frequency The average power spectrum at that location; E The energy of the current target acoustic signal; N This represents the total number of sampling points for the target acoustic signal. For the target acoustic signal The amplitude of each sampling point; This is the energy correction factor; is the reference signal energy corresponding to the minimum acousto-magnetic time difference, e is the natural constant; the noise estimation segment duration is fixed at 300ms, and the frame length is fixed at 100ms; the 10-fold noise power spectrum is the noise reduction coefficient of spectral subtraction; 0.9 and 0.8 are the optimized parameters of the correction coefficients.

[0076] Taking the detection of low-quality signals in an underground high-voltage cable tunnel as an example, the signal-to-noise ratio is 15dB, the maximum peak-to-peak value is 0.008, and the sampling rate is 16kHz. The system automatically enters this scheme and locks into operation. First, the signal 300ms before magnetic field triggering is selected as the noise estimation segment, divided into 3 frames with a frame length of 100ms. The standard deviations of each frame are calculated to be 0.002, 0.003, and 0.0025, respectively, with a minimum standard deviation of 0.002 and a median standard deviation of 0.0025. The frame with a difference of less than 0.00025 from the minimum standard deviation is identified as a noise frame. After calculating its average power spectrum, the target acoustic signal power is... The signal is enhanced by subtracting 10 times the average power spectrum from the original spectrum. The enhanced signal is then divided into three segments, and the cross-correlation coefficients of the three acquired signals are calculated for each segment. The minimum correlation coefficients for each segment are 0.3, 0.28, and 0.32, respectively, with a median of 0.3. Since there are no outliers where the difference between the minimum correlation coefficient and the median exceeds 0.03, all signals are retained as valid signals and averaged to improve the signal-to-noise ratio to 28 dB. Finally, the reference energy corresponding to the minimum acoustic-magnetic time difference is recorded as 0.8. The energy of the current target acoustic signal is calculated as 0.08 based on the sum of squares of the amplitudes at each sampling point. This energy is then substituted into the energy correction coefficient formula to calculate... The coefficient is approximately 1.506. This coefficient is used to correct the current acousto-magnetic time difference. Significant peaks are identified from the accumulated average signal and finally determined as effective trigger points, successfully solving the positioning problem under low signal-to-noise ratio and strong interference.

[0077] The preferred embodiments of this invention are adapted to low-quality signal scenarios, effectively suppressing background noise and eliminating abnormal data packets caused by strong interference, significantly improving the purity and integrity of low signal-to-noise ratio signals; cumulative averaging processing can iteratively optimize signal quality, and energy adaptive correction can accurately compensate for positioning deviations caused by differences in propagation paths and media; combined with mode-locking design, it avoids frequent scheme switching, successfully breaking through technical bottlenecks such as low signal-to-noise ratio, signal distortion, and strong interference, and greatly improving the accuracy, reliability, and adaptability to complex environments of high-voltage cable partial discharge positioning.

[0078] Preferably, the step of calculating the current acoustic signal energy based on the acoustic signal segment specifically involves: calculating the sum of squares of the amplitudes of all sampling points in the acoustic signal segment, and using the sum of squares as the current acoustic signal energy.

[0079] More preferably, the reference energy is the acoustic signal energy corresponding to the smallest acoustic-magnetic time difference in historical detection data.

[0080] Among them, acoustic signal energy refers to the core parameter used for subsequent energy coefficient correction, which is a quantitative representation of the intensity of the sound signal segment; the sound signal segment refers to the target acoustic signal segment used for energy calculation after preprocessing in the partial discharge detection of high-voltage cables; the reference energy refers to the acoustic signal energy corresponding to the minimum acoustic-magnetic time difference in historical detection data, which serves as the benchmark for energy correction.

[0081] Specifically, the acoustic signal energy is calculated by summing the squares of the amplitudes of all sampling points in the sound signal segment. This sum is directly used as the acoustic signal energy for this detection. The calculation logic is simple and accurately reflects the signal strength, providing a reliable quantitative basis for energy correction. The reference energy is defined by selecting the record with the smallest acoustic-magnetic time difference from the historically accumulated valid data, extracting the corresponding acoustic signal energy, and setting it as the reference energy. This ensures a unified and effective benchmark for energy correction, guaranteeing the accuracy of the correction results.

[0082] In summary, the acoustic-magnetic time difference calculation method based on signal quality adaptation provided by this invention first completes signal interception, filtering, resampling, and extreme value purification through standardized signal preprocessing, effectively eliminating invalid noise and providing a clean signal foundation for subsequent positioning. Then, based on a three-level positioning scheme with dynamic matching of signal-to-noise ratio and peak-to-peak value as dual parameters, high-quality signals use direct peak positioning to ensure efficiency and accuracy, medium-quality signals use similar extreme value matching to suppress noise interference, and low-quality signals use spectral enhancement, segmented cross-correlation screening, and cumulative averaging noise reduction to gradually optimize signal quality. Moreover, the low-quality mode is locked after startup to avoid frequent switching. Finally, using the acoustic signal energy corresponding to the minimum acoustic-magnetic time difference in historical detection data as a reference, the deviation caused by the difference in propagation path and medium is compensated through energy adaptive correction. This comprehensively overcomes the technical bottlenecks of low signal-to-noise ratio, strong interference, and signal distortion, significantly improving the accuracy, reliability, and adaptability to complex environments of high-voltage cable partial discharge positioning, and achieving precise adaptation to all scenarios from ideal laboratories to harsh field sites.

[0083] The above description is merely a preferred embodiment of the technical solution of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for calculating the acousto-magnetic time difference based on signal quality adaptation, characterized in that, Includes the following steps: Signal acquisition and triggering benchmark establishment steps: acquire the acoustic and electromagnetic signals generated by partial discharge of the high-voltage cable, and perform trigger detection on the electromagnetic signals to determine the triggering time of the electromagnetic signals; Signal preprocessing and quality assessment steps: Based on the triggering time of the electromagnetic signal, extract a sound signal segment and a background noise signal segment of a preset time length; The audio signal segment and the background noise signal segment are preprocessed, and the signal-to-noise ratio and maximum peak-to-peak value are calculated based on the preprocessed signal. The adaptive scheme selection steps are as follows: The calculated signal-to-noise ratio (SNR) and maximum peak-to-peak value (MPV) are compared with preset threshold conditions. Based on the comparison results, the corresponding trigger point localization scheme is dynamically selected. The preset threshold conditions include a first SNR threshold, a second SNR threshold, a first MPV threshold, and a second MPV threshold. The trigger point localization schemes include a direct peak localization scheme, a similar extreme value matching localization scheme, and a piecewise correlation and cumulative averaging localization scheme. The dynamic selection of the corresponding trigger point localization scheme based on the comparison results includes: if the SNR is greater than the first SNR threshold and the maximum MPV threshold is greater than the first MPV threshold, then the direct peak localization scheme is selected; if the SNR is not greater than the first SNR threshold but is greater than the second SNR threshold, and the maximum MPV threshold is greater than the second MPV threshold, then the similar extreme value matching localization scheme is selected; otherwise, the piecewise correlation and cumulative averaging localization scheme is selected. The similar extreme value matching and localization scheme includes: an extreme value matching step: performing distribution matching between the extreme points of the acquired sound signal and the reference sound signal; wherein, the distribution matching includes: matching the minimum points of the two and retaining the minimum point pairs with positional deviations within a preset range, locating the maximum points between adjacent minimum points based on the minimum point pairs, constructing their respective extreme value sequences, calculating the positional difference between the corresponding maximum points and adjacent minimum points in the two sets of sequences, and retaining the maximum points with positional difference deviations within a preset range as the matched extreme value pairs; a gradient judgment step: calculating the bilateral gradient symmetry score for the candidate peak points in the matched extreme value pairs; and a trigger point judgment step: determining the first candidate peak point with a bilateral gradient symmetry score exceeding a preset score threshold as a valid trigger point. The segmented correlation and cumulative averaging localization scheme includes: a signal enhancement step: performing spectral subtraction enhancement processing on the sound signal segment based on the background noise signal segment; a segmented filtering step: performing segmented cross-correlation analysis on the enhanced multiple sound signals, calculating the segmented cross-correlation coefficients between the multiple sound signals, and removing abnormal signal segments with minimum correlation coefficients lower than the median and differences exceeding a preset difference to obtain the filtered signal segments; and a cumulative averaging step: performing cumulative averaging processing on the filtered signal segments, and determining the effective trigger point from the cumulatively averaged signal. Trigger point location steps: Determine the effective trigger point of the sound signal based on the selected location scheme; Time difference calculation and correction steps: Calculate the acoustic signal energy acquired this time based on the sound signal segment; calculate the original acoustomagnetic time difference based on the effective trigger point and the trigger point of the electromagnetic signal, and combine the sound signal energy with the preset reference energy to perform energy adaptive correction on the original acoustomagnetic time difference to obtain the final acoustomagnetic time difference.

2. The method for calculating the acoustic-magnetic time difference according to claim 1, characterized in that, The trigger detection includes: detecting whether the amplitude of the electromagnetic signal exceeds a preset fixed threshold, and determining the moment when the amplitude first exceeds the fixed threshold as the trigger moment of the electromagnetic signal.

3. The method for calculating the acoustic-magnetic time difference according to claim 1, characterized in that, The direct peak positioning scheme includes: In the sound signal segment, the first peak point whose amplitude exceeds a preset multiple of the maximum peak-to-peak value is determined as the valid trigger point.

4. The method for calculating the acoustic-magnetic time difference according to claim 1, characterized in that, The calculation process for the two-sided gradient symmetry score includes: Multiple sampling points are taken before and after the candidate peak point, and the sign consistency ratio of the forward gradient and the backward gradient is calculated. The ratio is used as the bilateral gradient symmetry score.

5. The method for calculating the acoustic-magnetic time difference according to claim 1, characterized in that, The spectral subtraction and enhancement processing includes: The background noise signal segment is divided into segments with a preset frame length. Stable noise frames are selected, and the noise power spectrum is estimated based on the noise frames. A fixed multiple of the noise power spectrum is then subtracted from the power spectrum of the audio signal segment to achieve signal enhancement.

6. The method for calculating the acoustic-magnetic time difference according to claim 1, characterized in that, The calculation of the acoustic signal energy based on the acoustic signal segment is specifically as follows: Calculate the sum of squares of the amplitudes of all sampling points in the sound signal segment, and use the sum of squares as the sound signal energy for this time.

7. The method for calculating the acoustic-magnetic time difference according to claim 1, characterized in that, The reference energy is the acoustic signal energy corresponding to the smallest acousto-magnetic time difference in historical detection data.

Images