Medium and high voltage cable insulation defect live detection and positioning method

Abstract

The invention relates to the technical field of power cable fault detection and positioning, in particular to a middle and high voltage cable insulation defect live detection and positioning method, which comprises the following steps: step 1, injecting a composite detection signal into a cable conductor through a signal injection device; 2, arranging signal acquisition devices at a plurality of preset point positions of the cable; step 3, carrying out adaptive processing on the acquired response signal; step 4, extracting insulation defect features from the response signals after adaptive processing; step 5, based on the extracted insulation defect features, classifying defect types by using a pattern recognition algorithm, and calculating defect positions based on a signal propagation time difference principle; and 6, outputting a defect identification and positioning result to a monitoring system, verifying, and if the deviation between the positioning result and historical data exceeds a threshold value, re-executing the steps 1 to 5. By combining a multi-point synchronous measurement technology and a precise space-time analysis method, the position where the defect occurs can be positioned more accurately.

Description

Technical Field

[0001] This invention relates to the field of power cable fault detection and location technology, and in particular to a method for live detection and location of insulation defects in medium and high voltage cables. Background Technology

[0002] Medium and high voltage cables are critical components of power transmission systems, and their insulation condition directly affects the reliability and safety of the power grid. In actual operation, cable insulation may develop defects due to aging, mechanical damage, and environmental factors, such as partial discharge, insulation layer cracks, or moisture intrusion. If these defects are not detected and located in a timely manner, they may lead to cable failures, power outages, and even equipment damage and safety risks. Therefore, developing effective methods for detecting and locating cable insulation defects is crucial for power system maintenance.

[0003] Methods for detecting cable insulation defects mainly fall into two categories: offline testing and online monitoring. Offline testing methods require the cable to be de-energized, such as assessing the insulation condition through DC withstand voltage tests or dielectric loss measurements. While this method offers high accuracy, it necessitates power outages, disrupting power supply continuity, and cannot reflect the cable's operational status in real time. Online monitoring methods, on the other hand, allow detection while the cable is energized. Common methods include time-domain reflectometry (TDRS), frequency-domain reflectometry (FDR), and partial discharge detection. TDRS identifies defects by injecting pulse signals into the cable and analyzing the reflected signals. However, this method is susceptible to electromagnetic interference under high-voltage conditions, leading to signal attenuation and distortion, making it difficult to accurately identify minute defects. FDR analyzes cable impedance changes by scanning a frequency range, but its resolution is limited by the frequency bandwidth, and signal processing is complex under high-voltage conditions, resulting in significant location errors. Partial discharge detection identifies insulation defects by monitoring partial discharge signals in the cable. However, this method typically requires multiple sensors working together, leading to high costs and susceptibility to environmental noise, resulting in false alarms or missed detections.

[0004] Furthermore, existing online monitoring methods have limitations in defect localization. For example, time-domain reflectometry and frequency-domain reflectometry rely on accurate calculation of signal propagation speed, but the cable insulation material and operating conditions can affect signal speed, leading to location errors. While partial discharge detection can identify defect types, its localization accuracy is affected by sensor placement and signal propagation path, making high-precision localization difficult. Additionally, existing methods often lack adaptability to specific application scenarios of medium- and high-voltage cables. For instance, in long-distance cables or complex power grid structures, signal attenuation and interference are more severe, making it difficult for existing technologies to effectively distinguish defect signals from noise.

[0005] In summary, existing technologies for live-line detection and location of insulation defects in medium and high voltage cables suffer from the following main problems: First, the detection signal is easily interfered with by the high-voltage environment, resulting in a low signal-to-noise ratio; second, the defect location accuracy is insufficient, making it difficult to meet the needs of precise maintenance; and third, the methods are complex and costly, hindering large-scale application. Therefore, there is an urgent need for a highly innovative, adaptable, and cost-effective live-line detection and location method to address these issues. Summary of the Invention

[0006] To achieve the above objectives, the present invention provides a method for live detection and location of insulation defects in medium and high voltage cables, comprising the following steps:

[0007] Step 1: When the cable is energized, a composite detection signal is injected into the cable conductor through a signal injection device. The composite detection signal consists of multiple frequency components. The selection of frequency components is determined based on the cable's insulation material and operating voltage level. The amplitude of the composite detection signal is dynamically adjusted according to the cable's operating voltage.

[0008] Step 2: Deploy signal acquisition devices at multiple predetermined points along the cable to acquire the response signals generated by the propagation of the composite detection signal in the cable. The predetermined points are evenly distributed along the length of the cable. The signal acquisition devices include voltage sensors and current sensors, which simultaneously acquire voltage response signals and current response signals. After acquisition, preliminary filtering processing is performed.

[0009] Step 3: Adaptive processing is performed on the acquired response signal. Adaptive processing includes time-frequency analysis, energy distribution identification, adaptive filtering, and normalization to enhance defect-related features and suppress noise.

[0010] Step 4: Extract insulation defect features from the adaptively processed response signal. Insulation defect features include signal distortion factor, phase offset, and energy attenuation rate.

[0011] Step 5: Based on the extracted insulation defect features, the defect type is classified using a pattern recognition algorithm, and the defect location is calculated based on the signal propagation time difference principle. The signal propagation time difference is obtained by comparing the arrival time difference of response signals from different channels.

[0012] Step 6: Output the defect identification and location results to the monitoring system and verify them. If the location results deviate from the historical data by more than a threshold, repeat steps 1 to 5.

[0013] Preferably, the specific process of generating the composite detection signal in step 1 includes:

[0014] The frequency components cover a range from low to high frequencies. Low-frequency components are used to penetrate the cable insulation layer and reflect the overall insulation condition, while high-frequency components are used to capture local defect characteristics. The specific values ​​of the frequency components are determined through pre-testing. The pre-testing process involves measuring the cable's impedance spectrum under normal operating conditions and selecting the frequency component based on the resonant point in the impedance spectrum to ensure effective signal propagation in the cable and avoid resonance interference. The amplitude adjustment process for the composite detection signal involves setting the signal amplitude according to the percentage of the cable's operating voltage to avoid affecting the normal operation of the cable. The signal injection device uses a coupling transformer to achieve signal injection, ensuring electrical isolation from the high-voltage cable. The parameters of the coupling transformer are selected according to the cable voltage level and signal frequency range to match the cable impedance.

[0015] Preferably, the specific process of acquiring the multi-channel response signal in step 2 includes:

[0016] The spacing between predetermined points is determined based on the total cable length and expected positioning accuracy. The calculation process for the point spacing is as follows: divide the total cable length by the number of predetermined points minus one. The number of predetermined points is set based on the cable length and the minimum detectable defect size. The sampling rate setting process for the signal acquisition device is as follows: based on the highest frequency component of the composite detection signal, the sampling rate is set to more than twice that of the highest frequency component to ensure signal integrity. The preliminary filtering process uses a bandpass filter. The passband range of the bandpass filter is consistent with the frequency range of the composite detection signal. The design of the bandpass filter is based on the typical noise spectrum of the cable, which is obtained through historical operating data. The calibration process for the voltage and current sensors is as follows: with the cable in a defect-free state, a reference signal is acquired, and the sensor gain and offset are adjusted based on the reference signal.

[0017] Preferably, the specific sub-steps of adaptive processing of the response signal in step 3 include:

[0018] Time-frequency analysis uses short-time Fourier transform (SFT) to convert the signal into a time-frequency domain representation. The window function type and window length of the SFT are dynamically selected based on signal stability, which is evaluated through signal variance. The energy distribution identification process involves calculating the energy value of the time-frequency domain signal and setting an energy threshold. The energy threshold is determined based on historical energy data under normal cable operation conditions, and regions with energy values ​​exceeding the energy threshold are identified as abnormal regions. The parameters of the adaptive filter are dynamically adjusted according to the cable's operating status, including cable load current and ambient temperature. Cable load current is acquired in real time using a current sensor, and ambient temperature is acquired in real time using a temperature sensor. The update rate of the adaptive filter is set based on the signal change rate, which is calculated using the signal derivative. Normalization is performed based on the root mean square (RMS) value of the signal, which is calculated using a sliding window, with the window size adjusted according to the signal frequency components.

[0019] Preferably, the specific process of extracting insulation defect features in step 4 includes:

[0020] The signal distortion factor is obtained by calculating the difference between the response signal and the reference signal. The reference signal is the simulated signal or historical data of the cable under defect-free conditions. The simulated signal is generated based on the cable's geometric parameters and material properties. The difference is calculated using the root mean square error method. The phase offset is calculated by comparing the phase difference of the response signals from different channels. The phase difference is calculated using a cross-correlation algorithm, and the window length of the cross-correlation algorithm is set according to the signal period. The energy attenuation rate is obtained by analyzing the attenuation characteristics of the response signal in the frequency domain. The attenuation characteristics are determined by fitting the relationship curve between frequency and amplitude. The fitting process uses the least squares method, and the weights of the least squares method are allocated based on the signal-to-noise ratio. After feature extraction, the feature values ​​are standardized by dividing the feature value by a reference value related to the cable length and type. The reference value is obtained from the cable specification or test data.

[0021] Preferably, the specific process of insulation defect identification and location calculation in step 5 includes:

[0022] The pattern recognition algorithm employs a support vector machine (SVM). The SVM classification model is trained using historical defect data, which includes feature samples of various defect types. The training process includes feature selection and hyperparameter optimization. Defect types include partial discharge, insulation aging, and mechanical damage. The distinction between defect types is based on a combined threshold of feature values, which is obtained through learning from the training data. In the location calculation, the signal propagation speed is calibrated using cable parameters and operating conditions. Cable parameters include insulation dielectric constant and conductor resistance, while operating conditions include temperature and voltage levels. The calibration process involves injecting a test signal at a known location on the cable, measuring the signal propagation time, and then calculating the propagation speed. The time difference of arrival is calculated using a cross-correlation function, and the peak value of the cross-correlation function is detected using an interpolation method to improve accuracy. The defect location is determined using a weighted average method, with weights allocated based on signal quality. Signal quality is evaluated using the signal-to-noise ratio (SNR), which is calculated as the ratio of signal power to noise power.

[0023] Preferably, the specific process of result output and verification in step 6 includes:

[0024] The verification process includes comparison with historical data and on-site testing confirmation. Historical data comes from previous cable inspection records. The comparison process calculates position deviation and type consistency. The threshold is set according to the cable length and positioning accuracy requirements, and the threshold is calculated as a percentage of the cable length. If steps 1 to 5 are re-executed, the frequency components and amplitude of the composite detection signal are adjusted, and the signal quality feedback based on the previous detection is adjusted. The signal quality feedback includes the signal-to-noise ratio and feature stability index. The output results include defect location coordinates, defect type, and confidence level. The confidence level is calculated using the output probability of a pattern recognition algorithm.

[0025] Preferably, the specific configuration of the signal injection device in step 1 includes:

[0026] The composite signal generator produces a composite detection signal. The output impedance of the composite signal generator matches the characteristic impedance of the cable. The matching process is achieved through impedance testing and adjustment circuits. The core material of the coupling transformer is selected based on the signal frequency to minimize magnetic loss. The signal injection point is located at the beginning or end of the cable. The injection point selection is based on the cable network topology, and the topology data is obtained from the power grid management system. The signal injection timing is set according to the cable load cycle to avoid interference during peak load periods.

[0027] Preferably, the impedance spectrum measurement process in the pre-test includes:

[0028] A sweep frequency generator is used to inject a sweep frequency signal into the cable, and the sweep frequency range covers the frequency range of the composite detection signal. The reflection coefficient and transmission coefficient of the cable are measured and obtained by a vector network analyzer. The resonant point is identified as the phase change point in the impedance spectrum, and the phase change point is detected by phase derivative. The frequency components are selected to avoid harmonic interference, and harmonic interference is identified by spectrum analysis.

[0029] Preferably, the sensor arrangement and data processing of the signal acquisition device include:

[0030] The voltage sensor uses a capacitive voltage divider, and the voltage division ratio of the capacitive voltage divider is calibrated according to the cable voltage level; the current sensor uses a Rogowski coil, and the sensitivity of the Rogowski coil is tested by a standard current source; the data acquisition unit synchronizes the sampling time of multiple channels, and the synchronization signal is provided by a GPS module or fiber optic network; after preliminary filtering, the signal is compressed to reduce the amount of data. The compression algorithm is based on wavelet transform, and the wavelet basis function is selected according to the signal characteristics.

[0031] The beneficial effects of this invention are:

[0032] 1. This invention significantly improves signal quality and signal-to-noise ratio under high-voltage environments by optimizing signal injection and reception methods and employing advanced noise filtering and signal enhancement technologies. Through a specially designed cable insulation monitoring system, the impact of electromagnetic interference on the detection signal can be reduced, ensuring accurate detection of even minute insulation defects while the circuit is energized.

[0033] 2. This invention, by combining multi-point synchronous measurement technology and precise spatiotemporal analysis methods, can more accurately locate the position of defects. This invention can automatically adjust and optimize signal processing algorithms to adapt to different cable materials and environmental conditions, greatly improving the accuracy of defect location and meeting the needs of precise maintenance.

[0034] 3. This invention employs a simplified detection device and low-cost sensors, while integrating advanced data processing technology, making the entire system more economical and practical. Compared with traditional methods, this invention not only reduces equipment costs but also maintenance expenses, and the system is easy to operate, making it suitable for large-scale application. Through modular design, users can select different configurations according to their actual needs, further improving the system's applicability and cost-effectiveness.

[0035] 4. This invention utilizes adaptive signal processing technology designed for different cable types and operating environments to effectively address signal attenuation and interference issues in long-distance cables and complex power grid structures. The system can dynamically adjust detection parameters based on the real-time status of the cable, achieving efficient monitoring of cable insulation status and ensuring reliability in complex application scenarios. Attached Figure Description

[0036] To more clearly illustrate the technical solutions in this invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, those skilled in the art can obtain other drawings based on these drawings without creative effort.

[0037] Figure 1 This is a flowchart of the steps of the method of the present invention;

[0038] Figure 2 This is a flowchart of the steps for generating the composite detection signal in step 1 of the method of the present invention;

[0039] Figure 3 This is a flowchart of the multi-channel response signal acquisition steps in step 2 of the method of the present invention. Detailed Implementation

[0040] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. It should also be noted that, to make the embodiments more comprehensive, the following embodiments are the best and preferred embodiments, and those skilled in the art can use other alternative methods to implement some well-known technologies; moreover, the accompanying drawings are only for more specific description of the embodiments and are not intended to specifically limit the present invention.

[0041] Please see Figures 1-3This invention provides a method for live detection and location of insulation defects in medium and high voltage cables. In step 1, a composite detection signal is first injected into the cable conductor using a signal injection device while the cable is energized. This composite detection signal consists of multiple frequency components. The selection of different frequency components is adjusted according to the type of insulation material and the operating voltage level of the cable to ensure that the signal adapts to the electrical characteristics and operating environment of the cable. The amplitude of the composite signal is adjusted according to the dynamic changes in the actual operating voltage of the cable to maintain the stability and adaptability of the detection signal. Detection can be performed while the cable is energized, avoiding power outages and thus not affecting the continuity of power supply. Furthermore, the multi-frequency signal improves the detection sensitivity of defects.

[0042] In step 2, signal acquisition devices are deployed at multiple predetermined points along the cable's length, with these points evenly distributed. The signal acquisition devices include voltage and current sensors, which simultaneously acquire voltage and current response signals propagating within the cable. After acquisition, preliminary filtering is performed to remove stray signals and noise. This accurately captures the response signals within the cable, and by simultaneously acquiring voltage and current signals, a comprehensive understanding of the cable's operating status is gained, thereby improving detection accuracy.

[0043] In step 3, the acquired response signal undergoes adaptive processing, including time-frequency analysis, energy distribution identification, adaptive filtering, and normalization. These processing methods effectively enhance defect-related feature signals and suppress background noise, ensuring that defect feature signals are more prominent. Time-frequency analysis allows for the analysis of signal variations in the time and frequency domains, thereby accurately identifying anomalies in the cable. It enables the extraction of accurate defect features from complex electrical environments, thus avoiding the effects of signal attenuation and noise interference.

[0044] In step 4, insulation defect features are extracted from the adaptively processed response signal, mainly including signal distortion factor, phase offset, and energy attenuation rate. These features help identify the types of insulation defects that may exist in the cable and provide basic data for subsequent defect classification and location. Through in-depth analysis of the signal, different types of defects can be accurately identified, especially small or hidden defects, significantly improving the reliability of detection.

[0045] In step 5, based on the extracted insulation defect features, a pattern recognition algorithm is used to classify the defect types, and the defect location is calculated using the signal propagation time difference principle. By comparing the arrival time differences of the response signals at different acquisition points, the specific location of the defect can be determined. Utilizing the signal propagation time difference enables precise defect localization, allowing for the rapid identification of insulation defects in the cable and timely implementation of repair measures.

[0046] In step 6, the results of defect identification and location are output to the monitoring system for verification. If the location result deviates from historical data by more than a set threshold, steps 1 to 5 are repeated for supplementary inspection. The monitoring system provides real-time feedback on the inspection results and compares them with historical data to further ensure the accuracy and reliability of the inspection results, avoiding incorrect location due to system errors or abnormal environments.

[0047] This invention, by employing multi-frequency composite signals, synchronous acquisition, precise adaptive signal processing, and a time-difference-based positioning algorithm, not only overcomes problems such as signal interference and insufficient positioning accuracy, but also possesses strong adaptability and high efficiency. It can provide accurate and real-time cable insulation defect detection and positioning services in complex electrical environments, and has broad application prospects.

[0048] In one possible implementation, the frequency components of the composite detection signal first cover a range from low to high frequencies. The low-frequency components penetrate the cable's insulation layer, reflecting the overall insulation condition of the cable, while the high-frequency components are specifically designed to capture the characteristics of localized defects. This frequency component design simultaneously meets the detection requirements for both overall insulation condition and localized defects, improving the comprehensiveness and accuracy of defect detection.

[0049] Secondly, the specific values ​​of the frequency components of the composite signal are determined through pre-testing. The pre-testing process is conducted under normal cable operating conditions. This involves measuring the cable's impedance spectrum and selecting appropriate frequency components based on the resonant points. This process ensures effective signal propagation within the cable while avoiding resonance between the frequency and the cable itself, preventing resonance interference from affecting signal transmission and thus guaranteeing the reliability of the test results.

[0050] In the process of adjusting the amplitude of the composite detection signal, the signal amplitude is set according to a percentage of the cable's operating voltage. This adjustment ensures that the signal strength is moderate, effectively transmitting detection information without affecting the normal operation of the cable. This amplitude adjustment method avoids electrical interference to the cable itself that may be caused by an excessively strong signal, effectively ensuring the stability of the power system.

[0051] Signal injection is achieved through a coupling transformer, ensuring electrical isolation between the injected signal and the high-voltage cable. The coupling transformer is selected based on parameter matching between the cable voltage level and the signal frequency range, thereby maximizing signal transmission efficiency and avoiding signal distortion or attenuation. This electrical isolation design prevents potential electrical shocks to the cable equipment, further ensuring its safe operation.

[0052] In summary, this invention effectively solves problems such as limited signal propagation, resonance interference, and signal strength mismatch by precisely selecting the frequency components of the composite detection signal, making reasonable amplitude adjustments, and using a coupling transformer for electrical isolation. This greatly improves the detection accuracy and positioning accuracy of insulation defects in medium and high voltage cables, while ensuring the safety and stability of the power system.

[0053] In one possible implementation, the spacing between the predetermined points is first set based on the total length of the cable and the expected positioning accuracy. The point spacing is calculated by dividing the total cable length by the number of predetermined points and subtracting 1. The number of predetermined points is set according to the cable length and the smallest detectable defect size, ensuring uniform coverage of the entire cable's inspection area and that the distance between each inspection point is sufficient to meet positioning accuracy requirements. By rationally planning the point distribution, insulation defects in the cable can be located more accurately, improving the spatial resolution of the inspection and avoiding the omission of minute defects.

[0054] Secondly, the sampling rate setting of the signal acquisition device needs to take into account the highest frequency component of the composite detection signal. The sampling rate is set to more than twice the highest frequency component to ensure the integrity of signal acquisition and avoid signal distortion or information loss due to insufficient sampling. This embodiment of the invention can effectively guarantee high-fidelity signal acquisition, thereby providing high-quality response data and further ensuring the accuracy of subsequent defect analysis.

[0055] Next, a bandpass filter is used for preliminary filtering, with its passband range matching the frequency range of the composite detection signal. The bandpass filter is designed based on the typical noise spectrum of the cable, obtained from historical operational data. This design effectively removes stray noise from the cable system, ensuring a clearer and more accurate filtered signal and avoiding interference from noise in defect identification and location. Therefore, this embodiment of the invention significantly improves the quality of the detection signal, thereby enhancing the accuracy and reliability of the detection system.

[0056] Finally, the voltage and current sensors were calibrated by acquiring reference signals under defect-free cable conditions. The sensor gain and offset were then adjusted based on these reference signals. This calibration process ensures that the sensors maintain optimal operating conditions throughout the detection process, avoiding measurement errors caused by sensor performance deviations. This process guarantees sensor accuracy, thus providing more precise data support for defect location.

[0057] This invention effectively improves the accuracy and reliability of the cable insulation defect detection system by precisely setting the distance between detection points, reasonably setting the sampling rate, optimizing filtering, and calibrating the sensors, thus ensuring the accuracy of the detection results and enhancing the safety and operational stability of the power system.

[0058] In one possible implementation, time-frequency analysis first uses a Short-Time Fourier Transform (STFT) to convert the signal into a time-frequency domain representation. During the STFT, the type and length of the window function are dynamically selected based on the signal's stability. Signal stability is assessed by calculating the signal variance; a lower variance indicates signal stability, while a higher variance suggests potential fluctuations or anomalies. Dynamically selecting the window function type and window length optimizes the time-frequency domain decomposition of the signal, making the processed signal more representative and facilitating further analysis. This embodiment of the invention can select the optimal processing method under different signal stability conditions, thereby improving the accuracy and efficiency of signal processing.

[0059] Secondly, the energy distribution identification process identifies abnormal areas by calculating the energy value of the time-frequency domain signal and setting an energy threshold. The energy threshold is determined based on historical energy data under normal cable operating conditions, allowing for the establishment of a reasonable standard for the actual operating conditions of the cable. When the energy value of the signal exceeds this set threshold, the area is identified as an abnormal area, potentially indicating insulation defects or other electrical faults. This process effectively filters out irrelevant signals, focusing on areas that may indicate defects, thus improving the accuracy of defect detection.

[0060] The adaptive filter dynamically adjusts its parameters based on the cable's operating status. The cable's operating status includes load current and ambient temperature. Load current is acquired in real-time using a current sensor, while ambient temperature is acquired using a temperature sensor. The filter's update rate is set based on the rate of signal change, calculated using the signal's derivative. This dynamic adjustment process automatically adjusts the filter parameters when cable operating conditions change, thereby more effectively removing noise and improving the signal-to-noise ratio. This embodiment of the invention can optimize based on the cable's real-time operating status and changes in the external environment, avoiding the limitations of traditional fixed-parameter filters.

[0061] Finally, normalization is performed based on the root mean square (RMS) value of the signal, which is calculated using a sliding window. The window size is adjusted according to the signal's frequency components. This normalization helps reduce the impact of signal amplitude variations, ensuring the signal maintains a consistent scale under different conditions. Calculating the RMS using a sliding window smooths signal variations, removes short-term fluctuations, and makes the overall signal more stable, facilitating subsequent defect analysis.

[0062] This invention utilizes techniques such as time-frequency analysis, energy distribution identification, adaptive filter dynamic adjustment, and normalization processing to effectively handle signals generated by insulation defects in live cables, improving the accuracy, stability, and adaptability of the detection process. The embodiments of this invention enable more precise detection and location of cable insulation defects and maintain good performance in varying cable operating environments.

[0063] In one possible implementation, the signal distortion factor is first calculated by comparing the difference between the response signal and a reference signal. The reference signal can be a simulated signal or historical data from a defect-free cable state, where the simulated signal is generated based on cable geometry and material properties modeling. The root mean square error (RMSE) method is used to quantify the difference when calculating the signal difference. RMSE effectively reflects the overall deviation between signals with high accuracy. By extracting the signal distortion factor, it is possible to assess whether the cable exhibits signal distortion due to insulation defects, thus providing important clues for defect location.

[0064] Secondly, the phase offset is calculated by comparing the phase difference between the response signals of different channels, using a cross-correlation algorithm. The window length of the cross-correlation algorithm is dynamically adjusted according to the signal period to ensure accurate calculation of the phase difference across different signal periods. Changes in phase difference are typically related to the location and type of cable insulation defects, helping to further pinpoint the defect area.

[0065] The energy attenuation rate is extracted by analyzing the attenuation characteristics of the response signal in the frequency domain. This process first fits the relationship curve between frequency and amplitude using the least squares method. The weights of the least squares method are allocated according to the signal-to-noise ratio (SNR), allowing focus on the main components of the signal even at low SNR levels, thus avoiding noise interference. By analyzing the energy attenuation rate, regions of energy loss in the cable due to insulation damage or aging can be identified, helping to determine the severity of the defects.

[0066] Finally, after feature extraction, all feature values ​​need to be standardized. Standardization is achieved by dividing each feature value by a baseline value related to the cable length and type, which can be obtained from the cable specifications or test data. Standardization unifies the feature values ​​of different cables into a standardized range, facilitating cross-cable comparisons and defect analysis. Standardization eliminates deviations caused by differences in cable dimensions, allowing for a reasonable comparison of defect characteristics across different cables and ensuring consistency in test results.

[0067] This invention extracts multiple features such as signal distortion factor, phase offset, and energy attenuation rate, and through standardization processing, makes the detection and location of cable insulation defects more accurate and reliable. This process can effectively distinguish between normal and abnormal signals, helping to identify insulation defects in cables and providing a scientific basis for subsequent repair and maintenance work, thereby ensuring the safe and stable operation of the power system.

[0068] In one possible implementation, the pattern recognition algorithm first employs a Support Vector Machine (SVM). This algorithm is trained using historical defect data, which includes feature samples of various defect types. Defect types include partial discharge, insulation aging, and mechanical damage. During training, the SVM classification model undergoes feature selection and hyperparameter optimization. Feature selection aims to extract the most distinctive features from a large dataset, while hyperparameter optimization adjusts model parameters to achieve optimal classification performance. During training, defect type differentiation is based on combined threshold values ​​of feature values, which are learned from historical training data. This training process ensures the model has a high recognition rate for various defect types and can accurately distinguish between different types of defects.

[0069] Next, in the defect location calculation, the signal propagation speed needs to be calibrated first. The calibration process takes into account the cable's physical parameters and operating conditions. Cable parameters include the insulation dielectric constant and conductor resistance, while operating conditions include temperature and voltage levels. By injecting a test signal into a known location on the cable and measuring the signal propagation time, the signal propagation speed can be calculated. This step effectively compensates for differences in signal propagation speed caused by variations in cable parameters and environmental conditions, ensuring the accuracy of the location calculation.

[0070] In the localization calculation process, the signal arrival time difference is calculated using a cross-correlation function, and an interpolation method is employed to improve the accuracy of peak detection in the cross-correlation function. The peak value of the cross-correlation function corresponds to the shortest path for signal propagation, thus allowing for precise determination of the signal arrival time difference. Finally, a weighted average method is used to determine the defect location. The weight allocation is based on signal quality, which is evaluated using the signal-to-noise ratio (SNR). The SNR is the ratio of signal power to noise power; a higher SNR indicates better signal quality and is therefore assigned a larger weight. This weighted average method effectively integrates the localization results from various channels, improving the accuracy and reliability of defect localization.

[0071] The pattern recognition algorithm using support vector machines can efficiently and accurately identify various types of insulation defects in cables. Furthermore, the application of precise signal propagation speed calibration, peak detection of cross-correlation functions, and weighted averaging methods can improve the accuracy and robustness of defect location. This invention provides a high-precision, high-reliability method for detecting and locating cable insulation defects, effectively avoiding the error accumulation problem in traditional methods and providing strong protection for the safe operation of power systems.

[0072] In one possible implementation, the verification process first involves comparing the test results with historical data and conducting on-site testing for confirmation. Historical data comes from previous cable inspection records, which include the types, locations, and repair details of past defects. During the comparison, the system calculates the deviation between the defect location in the test results and the defect location in the historical records, and verifies the consistency of the defect type. This process effectively checks the system's accuracy and provides a reliable reference for defect localization. The calculation of location deviation helps confirm whether the test results meet the expected accuracy requirements, while type consistency ensures that the identified defect type matches the actual problem.

[0073] Secondly, the system sets a threshold based on cable length and positioning accuracy requirements, with the threshold calculated as a percentage of cable length. This setting ensures that the positioning accuracy requirements can be adjusted accordingly for different cable lengths to meet the needs of practical applications. The dynamic adjustment of the threshold makes the method more flexible and accurate when handling cables of different specifications.

[0074] If the detection results require re-execution of steps 1 through 5, the frequency components and amplitude of the composite detection signal are adjusted based on the signal quality feedback from the previous detection. Signal quality feedback includes the signal-to-noise ratio (SNR) and characteristic stability index. The SNR is used to evaluate the clarity and reliability of the signal, while the characteristic stability index reflects changes in signal characteristics. By adjusting these signal parameters, the accuracy of subsequent detections can be improved, thereby ensuring the accuracy of defect location.

[0075] Finally, the output includes the defect location coordinates, defect type, and confidence score. The confidence score is calculated using the probability output by the pattern recognition algorithm and measures the reliability of the defect location. The confidence score helps users assess the reliability of the detection results, enabling the system to not only provide location and type information but also quantify the accuracy of this information, thus providing greater assurance for subsequent decision-making and maintenance.

[0076] Overall, this invention, through comparison with historical data, on-site testing, and adjustments to signal quality feedback during the output and verification process, effectively improves the accuracy, stability, and reliability of cable insulation defect detection results. The embodiments of this invention ensure the reliability and adaptability of the detection system, providing accurate defect detection and location services under different cable types and operating conditions, thereby effectively guaranteeing the safe operation of the power system.

[0077] In one possible implementation, a composite signal generator first generates a composite detection signal. This composite detection signal is composed of signals of different frequencies and amplitudes, capable of effectively penetrating the cable insulation layer and stimulating a response from internal defects in the cable. To ensure accurate and efficient signal transmission, the output impedance of the composite signal generator needs to be matched with the characteristic impedance of the cable. This matching process is achieved through impedance testing and adjustment circuitry. In practice, impedance matching minimizes signal reflection loss during transmission, ensuring efficient signal transmission into the cable and reducing unnecessary energy loss. This step ensures the quality of signal injection and improves the accuracy of subsequent defect detection.

[0078] Secondly, the selection of the core material of the coupling transformer is crucial to the signal transmission effect. The selection of the core material is determined based on the signal frequency, with the aim of minimizing magnetic losses. By using a suitable core material, magnetic energy loss can be reduced during signal transmission, maintaining signal strength and stability, thereby improving the sensitivity of cable insulation defect detection.

[0079] The selection of signal injection points is determined based on the cable network topology. Cable network topology data comes from the power grid management system, which analyzes the structure, configuration, and layout of the power grid to determine the most suitable injection points. This configuration ensures that signals are injected at appropriate locations on the cable, maximizing coverage of potential cable defects and avoiding uneven signal injection.

[0080] Finally, the timing of signal injection is also crucial. To avoid injection during peak cable load periods, the timing should be set according to the cable's load cycle. Under heavy cable loads, signal injection may be subject to significant interference, affecting signal quality and detection results. Therefore, by properly planning the timing of signal injection, current interference during high load periods can be avoided, ensuring signal clarity and stability.

[0081] This invention, through the rational configuration of each component of the signal injection device, including impedance matching of the composite signal generator, selection of coupling transformer core material, appropriate injection point setting for the cable network topology, and scientific arrangement of signal injection timing, effectively improves the quality and stability of signal transmission, providing more accurate and reliable data support for the detection and location of cable insulation defects. The embodiments of this invention can significantly improve the accuracy, anti-interference capability, and adaptability of the detection system, providing a strong guarantee for the safe operation of the power system.

[0082] In one possible implementation, the impedance spectrum measurement process during pre-testing mainly involves systematically measuring and analyzing the cable's impedance characteristics using a sweep frequency analyzer, vector network analyzer, and spectrum analysis. This provides precise signal support for subsequent detection and location of insulation defects in medium- and high-voltage cables. This process ensures the efficiency and accuracy of the detection process.

[0083] First, the frequency sweeper is used to inject a sweep signal into the cable, with the sweep range covering the frequency range of the composite detection signal. The sweep signal, which varies gradually within a certain frequency range, comprehensively evaluates the cable's reflection and transmission characteristics at different frequencies. This frequency range setting ensures that the injected signal matches the subsequent composite signal, providing a unified frequency reference and ensuring that the cable's characteristics are fully measured.

[0084] Next, the cable's reflection coefficient and transmission coefficient are measured using a vector network analyzer. The reflection coefficient describes the degree of signal reflection when it encounters mismatched impedance in the cable, while the transmission coefficient represents the signal's propagation effect within the cable. Obtaining these two parameters using a vector network analyzer allows for a detailed understanding of the cable's impedance characteristics at different frequencies, thus providing a signal quality reference for subsequent testing.

[0085] Identifying resonant points is a crucial step in impedance spectrum analysis during measurement. Resonant points typically manifest as phase abrupt changes in the impedance spectrum. These abrupt changes are drastic phase shifts caused by physical structure or defects in the cable, indicating a significant change in impedance at that frequency. By detecting the phase derivative, these abrupt changes can be accurately located, further identifying potential defects or mismatches within the cable.

[0086] Furthermore, the selection of frequency components is crucial to avoid harmonic interference. Harmonic interference refers to high-order frequency components generated during signal propagation due to nonlinear effects; these components can interfere with the accuracy of the detection signal. Spectral analysis can identify and filter out these harmonic components, thereby ensuring the purity of the test signal. Effectively avoiding harmonic interference can significantly improve signal reliability and the accuracy of measurement results.

[0087] The impedance spectrum measurement process in the pre-test, through the injection of a swept-frequency signal, measurement of reflection and transmission coefficients, detection of phase abrupt changes at the resonant point, and optimized selection of frequency components, comprehensively assesses the impedance characteristics of the cable, providing accurate data support for subsequent insulation defect detection in medium and high voltage cables. This process effectively improves the sensitivity and accuracy of the detection, and by optimizing signal quality and reducing interference, it provides crucial assurance for rapid fault location in cables.

[0088] In one possible implementation, the voltage sensor first employs a capacitive voltage divider as the primary measuring tool. The capacitive voltage divider converts a high-voltage signal into a lower voltage signal through the voltage division effect of capacitors, facilitating subsequent measurement and analysis. The voltage division ratio of the capacitive voltage divider needs to be calibrated according to the cable's voltage rating. This calibration ensures accurate voltage signal acquisition under different voltage conditions. This technical feature ensures high accuracy and reliability of voltage measurement results, providing precise voltage data for defect detection.

[0089] Secondly, the current sensor employs a Rogowski coil, a highly sensitive measuring device widely used in non-contact current measurement. The sensitivity of the Rogowski coil is tested and calibrated using a standard current source to ensure accurate capture of current signal variations across different current ranges. The advantage of this configuration lies in its high sensitivity, enabling the detection of minute current fluctuations in cables, thereby helping to identify insulation defects or other potential problems.

[0090] In terms of data acquisition, multiple channels of the data acquisition unit need to sample synchronously to ensure consistent recording of data at multiple measurement points. To achieve this synchronization, the synchronization signal at the sampling time is provided by a GPS module or a fiber optic network. The GPS module provides global positioning signals to ensure precise synchronization of sampling times, while the fiber optic network ensures minimal latency during data acquisition through high-speed transmission. This synchronization mechanism is crucial because it ensures the consistency of data from different channels, avoids data errors caused by time deviations, and ensures the accuracy of the final data.

[0091] In the data processing, preliminary filtering is first performed to remove high-frequency noise and useless information from the signal, thereby improving signal clarity. The filtered signal is then further compressed to reduce data volume and improve storage and transmission efficiency. The compression algorithm is based on wavelet transform, which has excellent signal local analysis capabilities and can effectively compress signal data without losing key information. The wavelet basis functions are selected based on the characteristics of the signal, maximizing the preservation of important features while effectively reducing redundant data. This process significantly improves data processing efficiency and ensures signal integrity even with massive amounts of data.

[0092] The sensor arrangement and data processing of the signal acquisition device employ techniques such as capacitive voltage dividers, current sensors, GPS synchronization signals, fiber optic networks, and wavelet transform to ensure high accuracy, real-time performance, and efficient data processing in cable insulation defect detection. This invention effectively improves the performance of the detection system, ensures the reliability of the detection results, and provides a solid foundation for further data analysis and defect localization.

[0093] Example

[0094] Taking a 10kV, 2-kilometer-long cross-linked polyethylene insulated cable as an example, this cable operates in an urban power distribution network with a load current range of 200-500A and an ambient temperature range of -10℃ to 40℃. This cable has been in operation for 5 years, and recent monitoring has revealed increased partial discharge signals, necessitating defect detection and location.

[0095] 1. Generation of composite detection signals;

[0096] First, a preliminary impedance spectrum test of the cable was performed. An impedance analyzer (Agilent 4294A) was used to inject a swept frequency signal at the beginning of the cable, with a frequency sweep range of 1kHz to 2MHz, and a total of 1000 scan points were set. The impedance amplitude-frequency characteristic and phase-frequency characteristic curves of the cable were measured.

[0097] Impedance spectrum analysis revealed three distinct resonance points: the first at 15kHz, corresponding to the series resonance of the cable's distributed capacitance and inductance; the second at 85kHz, corresponding to dielectric relaxation of the insulation layer; and the third at 950kHz, corresponding to resonance between the cable sheath and the ground loop. Based on this resonance point distribution, frequencies of 10kHz, 50kHz, 200kHz, and 800kHz, avoiding the resonant frequencies, were selected as the frequency components of the composite detection signal.

[0098] The amplitude of the composite detection signal is dynamically adjusted according to the cable operating voltage. The specific calculation formula is as follows: ;

[0099] in, The amplitude of each frequency component, The rated voltage of the cable is 10kV. For the corresponding frequency, The highest frequency component (800kHz). For safety, a factor of 0.001 is used. The calculated amplitudes of each frequency component are as follows: 8.94V for 10kHz, 4.0V for 50kHz, 2.0V for 200kHz, and 1.0V for 800kHz.

[0100] The signal injection device uses a high-frequency coupling transformer with a turns ratio of 1:100. The primary winding is connected to a composite signal generator, and the secondary winding is connected to the cable conductor through a coupling capacitor. The coupling capacitor has a value of 1000pF and a withstand voltage rating of 15kV. The signal injection point is selected at the terminal block at the beginning of the cable, and the injection timing is chosen during the early morning period when the load current is relatively stable (approximately 250A).

[0101] 2. Multi-channel response signal acquisition;

[0102] Five signal acquisition points were arranged along the length of the cable, at distances of 0m, 500m, 1000m, 1500m, and 2000m from the starting end. The spacing of 500m between the points was determined based on the total cable length and the minimum detectable defect size (set to 0.5m), satisfying the Nyquist sampling theorem's requirements for spatial sampling.

[0103] The signal acquisition device at each location includes:

[0104] Voltage sensor: A capacitive voltage divider sensor is used, with a voltage division ratio of 1000:1, a bandwidth of 10kHz-1MHz, and an accuracy class of 0.5.

[0105] Current sensor: It adopts a wideband Rogowski coil with a sensitivity of 1V / A, a bandwidth of 10kHz-1MHz, and a linearity error of less than 1%.

[0106] Data acquisition unit: Uses a 16-bit ADC with a sampling rate of 2MHz to meet the sampling requirements of 800kHz frequency components.

[0107] The data acquisition system uses GPS clock synchronization, with a time synchronization error of less than 100ns at each point. Before signal acquisition, sensor calibration is performed: with the cable free of defects, a standard test signal (amplitude 10V, frequency 50kHz) is injected, the output of each sensor is measured, and the gain and offset are adjusted to ensure the measurement error is less than 1%.

[0108] The initial filtering process uses an 8th-order Chebyshev bandpass filter with a passband range of 8kHz-1MHz and a stopband attenuation greater than 60dB. The filter parameters are determined based on the noise characteristics of the cable's operating environment, primarily suppressing 50Hz power frequency interference and its harmonics, as well as high-frequency radio interference.

[0109] 3. Adaptive processing of response signals;

[0110] Time-frequency analysis: Short-time Fourier transform is used, and a Hamming window is selected as the window function. The window length is dynamically adjusted according to signal stability. Signal stability is evaluated by calculating the variance of the signal within the sliding window, with a variance threshold set to 0.01. When the signal variance is less than the threshold, the window length is set to 10 periods; when the variance is greater than the threshold, the window length is reduced to 5 periods to improve time resolution.

[0111] Energy distribution identification: Calculate the energy value of the time-frequency domain signal; the energy threshold is set based on historical normal operation data. The specific calculation method is as follows:

[0112] ;

[0113] in, The historical energy average. This represents the historical energy standard deviation. Energy values ​​exceeding... The region was identified as an anomalous region. Adaptive filtering: An LMS adaptive filter was used, with a step size parameter... Dynamically adjust according to cable operating status:

[0114] ;

[0115] in, Load current (obtained in real time via a current sensor, unit: A). Ambient temperature (obtained via PT100 temperature sensor, unit: °C). This is a reference temperature (taken as 20℃). and These are the weighting coefficients, taken as 0.001 and 0.01 respectively.

[0116] Normalization: A sliding window root mean square normalization method is used. The window size is set according to the highest frequency component of the signal, taking the period length of the two highest frequency components. For the 800kHz component, the window length is 2.5μs.

[0117] 4. Insulation defect feature extraction;

[0118] Signal distortion factor calculation: using the weighted root mean square error method.

[0119] ;

[0120] in, This is the measured signal. For reference signal, The weighting coefficient is proportional to the signal-to-noise ratio of each frequency component. The reference signal comes from the propagation characteristic test data of the cable at the time of manufacture. Phase offset calculation: A windowed cross-correlation algorithm is used, with a window length of one signal period. First, a Hilbert transform is performed on the signal to obtain the analytic signal, then the peak position of the cross-correlation function is calculated, and the phase difference is calculated using the following formula:

[0121] ;

[0122] in, The peak time difference of the cross-correlation function. Given the signal period. Energy attenuation rate calculation: The frequency-amplitude curve is fitted using the least squares method, with the fitting weights based on the signal-to-noise ratio of each frequency component. Attenuation rate. Calculated using the following formula:

[0123] ;

[0124] The attenuation rate value was obtained by taking the logarithm and then using linear fitting. Characteristic standardization: Each characteristic value was divided by a reference value related to the cable length (2000m) and cross-sectional area (240mm²), which was obtained from the specifications of cable model YJV22-8.7 / 10-3×240.

[0125] 5. Insulation defect identification and location calculation;

[0126] Support Vector Machine (SVM) model: Employing a radial basis function kernel, the model is trained using 500 sets of historical defect data, including 200 sets of partial discharge, 150 sets of insulation aging, and 150 sets of mechanical damage samples. The training process includes the following steps:

[0127] Feature selection: The recursive feature elimination method is used to select the 6 most important features from 10 candidate features.

[0128] Hyperparameter optimization: The penalty factor C and kernel function parameter γ were optimized using a grid search method. The optimal parameters were C=10 and γ=0.1.

[0129] Model validation: Five-fold cross-validation was used, and the accuracy rate reached 95.3%.

[0130] Defect type differentiation threshold:

[0131] Partial discharge: signal distortion factor > 0.25 and energy decay rate < 0.01

[0132] Insulation aging: Phase offset > 15° and energy decay rate > 0.02

[0133] Mechanical damage: Signal distortion factor > 0.3 and phase shift > 20°

[0134] Signal propagation speed calibration: A test signal is injected at a known location on the cable (1000m), and the arrival time at each point is measured. The actual propagation speed is calculated to be 1.65 × 10⁻⁶. 8 m / s. The calibration formula takes into account the effects of temperature and voltage:

[0135] ;

[0136] in, The nominal propagation speed (1.68 × 10⁻⁶) 8 m / s), The temperature coefficient is 0.002 / ℃. The voltage coefficient is 0.0001 / kV. Time difference of arrival calculation: Cubic spline interpolation is used to improve the peak detection accuracy of the cross-correlation function; the interpolation factor is 100 times, achieving a time resolution of 5 ns. Defect localization: Weighted least squares method is used, with weights allocated according to the signal-to-noise ratio.

[0137] ;

[0138] The signal-to-noise ratio (SNR) is calculated as the ratio of signal power to noise power, with noise power measured during signal quiescence.

[0139] 6. Results output and verification;

[0140] The output includes:

[0141] Defect location: 1256.3m from the starting point;

[0142] Defect type: Partial discharge;

[0143] Confidence level: 92.7%;

[0144] Eigenvalues: signal distortion factor 0.28, phase offset 8.5°, energy attenuation rate 0.008;

[0145] Verification process: Compared with historical data, the previous detection result showed partial discharge at a distance of 1255.1m from the starting point, with a location deviation of 1.2m. The threshold was set to 0.1% (2m) of the cable length, and the current deviation is within the allowable range. On-site test confirmation: After excavation at the location, dendritic discharge traces were found in the cable insulation layer, consistent with the diagnostic results. Confidence calculation: Based on the output probability of the support vector machine, the calculation formula is as follows:

[0146] ;

[0147] Where P is precision and R is recall, obtained through statistics from the test set.

[0148] This embodiment effectively suppresses on-site electromagnetic interference and achieves accurate detection of minute defects through composite detection signal design and adaptive signal processing. The use of multi-feature fusion and machine learning algorithms improves the accuracy of defect classification. The positioning method based on propagation time difference, combined with velocity calibration, controls the positioning error to within 0.1%, meeting on-site maintenance requirements.

[0149] This invention encompasses any substitutions, modifications, equivalent methods, and solutions made within the spirit and scope of this invention. To provide the public with a thorough understanding of this invention, specific details are described in detail in the following preferred embodiments; however, those skilled in the art will fully understand the invention even without these details. Furthermore, to avoid unnecessary misunderstanding of the essence of this invention, well-known methods, processes, procedures, components, and circuits are not described in detail.

[0150] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for detecting and locating defects in the insulation of a medium- or high-voltage cable under power, characterized in that, The method comprises the following steps: Step 1: injecting a composite detection signal into the cable conductor through a signal injection device under the live operation state of the cable, the composite detection signal being composed of multiple frequency components, the selection of the frequency components being based on the insulating material and the operating voltage grade of the cable, and the amplitude of the composite detection signal being dynamically adjusted according to the operating voltage of the cable; Step 2: arranging signal acquisition devices at multiple predetermined points of the cable, collecting response signals generated by the propagation of the composite detection signal in the cable, the predetermined points being uniformly distributed along the length direction of the cable, the signal acquisition device comprising a voltage sensor and a current sensor, and the voltage response signal and the current response signal being synchronously collected and then subjected to preliminary filtering processing; Step 3: performing adaptive processing on the collected response signals, the adaptive processing comprising time-frequency analysis, energy distribution identification, adaptive filtering and normalization processing, so as to enhance the defect-related features and suppress noise; Step 4: extracting insulating defect features from the adaptively processed response signals, the insulating defect features including signal distortion factor, phase shift amount and energy attenuation rate; Step 5: classifying the defect types based on the extracted insulating defect features using a pattern recognition algorithm, and calculating the defect position based on the signal propagation time difference principle, the signal propagation time difference being obtained by comparing the time difference of arrival of the response signals of different channels; Step 6: outputting the defect identification and positioning results to a monitoring system and verifying the results, and if the positioning result deviates from the historical data by more than a threshold value, re-executing steps 1 to 5.

2. The method according to claim 1, characterized in that, The specific process of generating the composite detection signal in step 1 comprises: The frequency components cover a range from low frequency to high frequency, the low frequency components being used to penetrate the cable insulation layer and reflect the overall insulation state, and the high frequency components being used to capture local defect features; The specific values of the frequency components are determined through pre-testing, the pre-testing process being: measuring the impedance spectrum of the cable under the normal operation state of the cable, and selecting the frequency components according to the resonance points in the impedance spectrum, so as to ensure effective propagation of the signal in the cable and avoid resonance interference; The amplitude adjustment process of the composite detection signal is: setting the signal amplitude according to the percentage of the operating voltage of the cable, so as to avoid affecting the normal operation of the cable; The signal injection device uses a coupling transformer to realize signal injection, ensuring electrical isolation from the high-voltage cable; The parameters of the coupling transformer are selected according to the voltage grade of the cable and the signal frequency range, so as to match the cable impedance.

3. The method of claim 1, wherein the method further comprises: The specific process of collecting the multi-channel response signal in step 2 comprises: The point spacing of the predetermined points is determined according to the total length of the cable and the expected positioning accuracy, the calculation process of the point spacing being: dividing the total length of the cable by the number of predetermined points minus one, and the number of predetermined points being set based on the cable length and the minimum detectable defect size; The sampling rate setting process of the signal acquisition device is: according to the highest frequency component of the composite detection signal, the sampling rate is set to be more than twice the highest frequency component, so as to ensure signal integrity; The preliminary filtering processing uses a band-pass filter, the passband range of the band-pass filter being consistent with the frequency range of the composite detection signal, and the design of the band-pass filter being based on the typical noise spectrum of the cable, the noise spectrum being obtained through historical operation data; The calibration process of the voltage sensor and the current sensor is: collecting a reference signal under the condition that the cable is in a defect-free state, and adjusting the sensor gain and offset based on the reference signal.

4. The method of claim 1, wherein the method further comprises: The specific sub-steps of adaptive processing of the response signal in step 3 include: The time-frequency analysis converts the signal into a time-frequency domain representation using a short-time Fourier transform, and the type and length of the window function of the short-time Fourier transform are dynamically selected according to the signal stability, which is evaluated by the signal variance; The energy distribution identification process is: calculating the energy value of the time-frequency domain signal, and setting an energy threshold, which is determined based on the energy historical data under the normal operation state of the cable, and the region with energy value exceeding the energy threshold is identified as an abnormal region; The parameters of the adaptive filter are dynamically adjusted according to the operating state of the cable, including the cable load current and the ambient temperature, which are obtained in real time by the current sensor and the temperature sensor, respectively, and the update rate of the adaptive filter is set according to the signal change rate, which is calculated by the signal derivative; The normalization process is based on the root mean square value of the signal, which is calculated by a sliding window, and the window size is adjusted according to the signal frequency component.

5. The method of claim 1, wherein, The specific process of insulation defect feature extraction in step 4 includes: The signal distortion factor is obtained by calculating the difference between the response signal and the reference signal, and the reference signal is the simulation signal or historical data under the condition that the cable is in a defect-free state, and the simulation signal is generated based on the geometric parameters and material properties of the cable, and the difference calculation uses the root mean square error method; The phase shift is calculated by comparing the phase difference of the response signals of different channels, and the phase difference is calculated using the cross-correlation algorithm, and the window length of the cross-correlation algorithm is set according to the signal period; The energy attenuation rate is obtained by analyzing the attenuation characteristics of the response signal in the frequency domain, and the attenuation characteristics are determined by fitting the relationship curve between frequency and amplitude, and the fitting process uses the least squares method, and the weights of the least squares method are assigned based on the signal signal-to-noise ratio; After feature extraction, the feature values are standardized, and the standardization process is: dividing the feature values by the reference values related to the length and type of the cable, which are obtained from the cable specification book or test data.

6. The method of claim 1, wherein, The specific process of insulation defect identification and positioning calculation in step 5 includes: The pattern recognition algorithm uses a support vector machine, and the classification model of the support vector machine is trained by historical defect data, which includes feature samples of multiple defect types, and the training process includes feature selection and hyperparameter optimization; The defect types include partial discharge, insulation aging and mechanical damage, and the distinction of the defect types is based on the combined threshold of the feature values, which is obtained by learning from the training data; In the positioning calculation, the signal propagation speed is calibrated by the cable parameters and operating conditions, including the dielectric constant of the insulation and the resistance of the conductor, and the operating conditions include temperature and voltage level, and the calibration process is: injecting a test signal at a known position of the cable, measuring the signal propagation time, and calculating the propagation speed inversely; The time difference of arrival is calculated by the cross-correlation function, and the peak detection of the cross-correlation function uses an interpolation method to improve the accuracy. The defect location is determined by a weighted average method, the weights are assigned according to signal quality, the signal quality is evaluated by signal-to-noise ratio, and the signal-to-noise ratio is calculated as the ratio of signal power to noise power.

7. The method of claim 1, wherein the method further comprises: The specific process of outputting and verifying the results in step 6 includes: The verification process includes comparison with historical data and on-site test confirmation, the historical data comes from the past detection records of the cable, and the comparison process calculates the position deviation and type consistency; The threshold is set according to the cable length and positioning accuracy requirements, and the threshold is calculated as a percentage of the cable length; If steps 1 to 5 are re-executed, adjust the frequency components and amplitudes of the composite detection signal, adjust the signal quality feedback based on the previous detection, the signal quality feedback includes signal-to-noise ratio and feature stability index; The output results include defect location coordinates, defect type and confidence, and the confidence is calculated by the output probability of the pattern recognition algorithm.

8. The method of claim 1, wherein the method further comprises: The specific configuration of the signal injection device in step 1 includes: The composite signal generator generates a composite detection signal, and the output impedance of the composite signal generator matches the characteristic impedance of the cable, and the matching process is realized by impedance testing and adjusting circuit; The core material of the coupling transformer is selected based on the signal frequency to minimize magnetic loss; The signal injection point is located at the start or end of the cable, and the injection point is selected based on the cable network topology, and the topology data is obtained from the power grid management system; The signal injection opportunity is set according to the cable load cycle to avoid interference during peak load period.

9. The method of claim 2, wherein the method further comprises: The measurement process of the impedance spectrum in the pre-test includes: A frequency sweep instrument is used to inject a frequency sweep signal into the cable, and the frequency sweep range covers the frequency range of the composite detection signal; The reflection coefficient and transmission coefficient of the cable are measured, and the reflection coefficient and transmission coefficient are obtained by a vector network analyzer; The resonance point is identified as a phase mutation point in the impedance spectrum, and the phase mutation point is detected by the phase derivative; The frequency component selection avoids harmonic interference, and the harmonic interference is identified by spectrum analysis.

10. The method of claim 3, wherein the method further comprises: The sensor arrangement and data processing process of the signal acquisition device includes: The voltage sensor uses a capacitive voltage divider, and the voltage division ratio of the capacitive voltage divider is calibrated according to the voltage grade of the cable; The current sensor uses a Rogowski coil, and the sensitivity of the Rogowski coil is tested by a standard current source; The data acquisition unit synchronizes the sampling time of multiple channels, and the synchronization signal is provided by a GPS module or a fiber optic network; After preliminary filtering, signal compression is performed to reduce data volume, and the compression algorithm is based on wavelet transform, and the wavelet basis function is selected according to the signal characteristics.

Images