Method for evaluating insulation aging state of cable

Abstract

The invention discloses a cable insulation aging state evaluation method, which realizes aging state evaluation of a running cable system by fusing data acquired by multiple sensors and combining an intelligent algorithm. The method comprises the following steps: 1, collecting cable operation data of different years, and constructing a preliminary data set; 2, generating simulation data to supplement the insufficiency of actually acquired data, and constructing a complete data set; 3, performing feature extraction on the signals in the complete data set; 4, carrying out weight analysis on the extracted features; 5, arranging a sensor, and collecting multi-source data in real time; 6, an alarm threshold value is set, early warning is carried out on the cable section with the aging score exceeding the threshold value, and the remaining life is evaluated; 7, synchronizing the early warning information to a maintenance management platform, and generating a detailed early warning report; and 8, checking an alarmed section, evaluating the alarm accuracy, and if the alarm is unreasonable, correcting the weight coefficient and the scoring criterion in time. By fusing data acquired by multiple sensors and combining an intelligent algorithm, the aging state of the operating system is evaluated.

Description

Technical Field

[0001] The present invention relates to the technical field of cable insulation aging status assessment, and in particular to a method for cable insulation aging status assessment, which realizes cable aging status assessment in operation by fusing multi-sensor data collection and combining intelligent algorithms. Background Art

[0002] High-voltage cables are essential for power transmission. Compared to oil-immersed and gas-insulated cables, cross-linked polyethylene (XLPE) power cables offer advantages such as simplified installation and easy maintenance. Currently, power grids with voltage levels of 10 to 220 kV primarily utilize XLPE cables as their primary insulation material. However, over long-term use, the insulation layer can be affected by various external factors, such as electric fields, heat, and mechanical stress, leading to insulation degradation. XLPE cables have numerous advantages, including the absence of oil supply equipment, excellent fire resistance, simple installation and maintenance, and superior mechanical and electrical performance, leading to their widespread use worldwide. However, XLPE cables have some significant drawbacks. Localized defects, such as insulation gaps and protrusions, can lead to localized electric field concentrations, which can easily cause insulation degradation phenomena such as water treeing. Furthermore, they can exhibit partial discharge (PD) over long-term operation.

[0003] In XLPE power cables, when affected by a single external cause or a combination of internal and external causes, insulation aging may take many forms, including thermal aging, mechanical aging, voltage aging, partial discharge aging, water tree aging, electrical tree aging, and chemical tree aging.

[0004] Cable insulation aging assessment methods can generally be divided into two categories. One is to evaluate the current insulation condition of the cable by testing single or multiple diagnostic parameters (i.e., test quantities), and perform diagnosis based on existing aging criteria. This can be further divided into diagnosis targeting the cable system and diagnosis targeting the cable insulation material. The other is to evaluate the remaining service life of the cable through accelerated aging tests and cable aging models.

[0005] However, most evaluation and diagnosis methods are destructive and cannot directly evaluate cables in operation, which makes it impossible to timely evaluate the aging status of cables in operation.

[0006] While measurements of dielectric loss factor, partial discharge parameters, and DC leakage current typically operate at low test voltages or require no voltage application, generally preventing insulation damage and constituting non-destructive testing, these dielectric parameters are not sufficiently sensitive to cable insulation degradation, making them difficult to reliably diagnose the aging of cable insulation. While the sensitivity of these tests could be further improved, specific insulation aging criteria have yet to be established for each diagnostic parameter to determine whether a cable should be replaced. Summary of the Invention

[0007] In response to the problems existing in the prior art, the purpose of the present invention is to propose a method for evaluating the aging status of cable insulation, which realizes the aging status evaluation of the running cable system by fusing multi-sensor data collection and combining intelligent algorithms.

[0008] In order to achieve the above object, the present invention adopts the following technical solutions:

[0009] A method for evaluating the aging status of cable insulation is provided. The method integrates data collected by multiple sensors and combines intelligent algorithms to evaluate the aging status of a running cable system. The method includes the following steps:

[0010] Step 1: Collect operating data of cables of different ages, correlate these actual operating data with the age of the cables, environmental conditions, and load conditions, and construct a preliminary data set;

[0011] Step 2: Generate simulation data to supplement the actual collected data and build a complete data set;

[0012] Step 3: Extract features from the signals in the complete data set to extract characteristic indicators that are valuable for cable aging and performance evaluation;

[0013] Step 4: Perform weight analysis on the extracted features;

[0014] Step 5: Deploy sensors to collect multi-source data in real time and monitor the cable system;

[0015] Step 6: Set an alarm threshold, issue an early warning for cable sections whose aging scores exceed the threshold, and evaluate the remaining lifespan;

[0016] Step 7: Synchronize the warning information to the maintenance management platform and generate a detailed warning report;

[0017] Step 8: Check the alarmed section to evaluate the alarm accuracy. If the alarm is unreasonable, promptly revise the weight coefficient and scoring criteria.

[0018] In step 2, the data set is supplemented by simulation technology. Simulation technology is used to simulate various possible cable aging scenarios, and their performance under different service years, operating conditions and fault conditions, generating a large amount of detailed simulation data. These simulation data, as a supplement, can effectively make up for the lack of field data and make the data set more comprehensive and diversified.

[0019] The specific process of step 3 is as follows: preprocess the collected sensor data, including data cleaning, denoising, and normalization, to ensure data quality and consistency; conduct in-depth data mining through time domain analysis, frequency domain analysis, and time-frequency analysis methods to extract key features.

[0020] The key features include the average value used to characterize the central tendency of the data, the variance used to measure the degree of dispersion of the data, the energy value used to characterize the energy distribution of the data, the energy entropy used to characterize the degree of disorder of the energy sequence of the data, the frequency entropy value used to characterize the complexity and disorder of each layer of sub-frequency band, and the entropy weight used to characterize the frequency entropy value of each layer and each time period as a percentage of the total frequency entropy value.

[0021] The specific process of step 4 is as follows: based on the feature extraction results and the constructed complete data set, a machine learning algorithm is used to perform weight analysis on the extracted features, calculate the contribution and correlation of each feature, determine the importance of the features, and sort them. After discarding features with lower weights, the remaining features are weighted and scored according to the weights. The feature weights are converted into specific scoring indicators, and a scoring criterion is constructed that associates feature values ​​with the degree of aging.

[0022] In step 4, the weight is calculated by the correlation between multi-source data and aging. Specifically, the cable operation data, environmental data and historical aging data are obtained from multiple sources to ensure the comprehensiveness and diversity of the data; the correlation between these multi-source data is explored, especially the correlation with the cable aging process; based on these correlations and the random forest algorithm, a scoring criterion for the correlation between characteristic values ​​and aging degree is constructed.

[0023] The specific process of step 5 is as follows: establish a sensor network to ensure that sensors can cover all key areas, connect each sensor to the data acquisition system, and use wired or wireless communication technology to achieve real-time transmission and centralized storage of data; perform real-time analysis and visualization of the collected multi-source data to provide real-time feedback on the operating status and health of the cable system.

[0024] The early warning report described in step 7 includes the specific location of the cable section, the aging score, and the parameters exceeding the threshold.

[0025] Compared with the prior art, the present invention has the following advantages:

[0026] 1. The data set is supplemented by simulation technology, and detailed simulation data is generated to make up for the lack of field data, making the data set more comprehensive and diversified, thereby improving the analysis accuracy and reliability of the early warning system; the system collects cable operation data, environmental data and historical aging data from multiple sources, and uses advanced data analysis technology to mine the correlation between multi-source data to construct a comprehensive aging scoring criterion, which can accurately reflect the aging behavior and future aging trend of cables under different conditions, thereby improving monitoring accuracy and early warning capabilities; the system uses the scoring criterion to score the operating cables in real time, and provides real-time health scores and early warnings by comprehensively considering factors such as the cable's immediate aging status and current load conditions, greatly enhancing the system's monitoring capabilities, enabling operation and maintenance personnel to obtain key status information in a timely manner, respond and make decisions quickly, and ensure the stable operation and safety of the system.

[0027] 2. Cable aging assessment using multi-source data collected by multiple sensors offers numerous advantages. Multi-source data collection enables a more comprehensive assessment, reflecting diverse aspects of the cable's operating status and avoiding potential misjudgments from a single data source. Multi-source data fusion also improves data accuracy and reliability. Correlation analysis of data from different sensors can more precisely pinpoint problem areas and the extent of aging. Furthermore, this rich data source enables the training of more complex and precise machine learning models, improving the accuracy and reliability of predictions and enhancing the system's early warning capabilities. Real-time monitoring of multi-source data allows for rapid detection of abnormal changes and prompting of early warning signals. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0029] Figure 1 A flow chart showing a method for evaluating the aging condition of cable insulation is shown. DETAILED DESCRIPTION

[0030] The purpose of the present invention is to realize the aging status assessment of the running system by fusing data collected by multiple sensors and combining intelligent algorithms.

[0031] To achieve the above object, the present invention provides a method for evaluating the aging state of cable insulation, the method comprising the following steps:

[0032] 1. Through the sensors installed in the cable system, various parameters in cable operation are continuously monitored, and cable operation data of different years are collected. These actual operation data are correlated with factors such as cable age, environmental conditions, and load conditions to build a preliminary data set.

[0033] 2. Based on the operating characteristics of the actual cable system and the physical properties of the cables, the actual cable line described in step 1 was constructed in the electromagnetic transient simulation software EMTP-ATP. The performance of the cable line under different service years, different operating conditions and fault states was analyzed. Simulation data was generated to supplement the actual collected data and construct a complete data set.

[0034] 3. Feature extraction is performed on the signals in the complete data set to identify valuable characteristic indicators for cable aging and performance assessment. The collected sensor data is preprocessed, including data cleaning, denoising, and normalization, to ensure data quality and consistency. The data is deeply mined through time domain analysis, frequency domain analysis, and time-frequency analysis to extract key features (including but not limited to mean, variance, energy entropy, frequency entropy, entropy weight, and other time-frequency domain features).

[0035] The method mainly includes two steps: 1) using stationary wavelet transform or Hilbert-Huang decomposition to obtain each sub-band of the fault signal; 2) dividing the frequency bands into equal lengths according to a certain interval based on the decomposition, and then extracting the feature value of each interval.

[0036] Taking the stationary wavelet transform as an example, assuming that the number of electrical quantities collected in each fault sample data is M and the total length of its time series data is N, each fault sample data can be expressed as:

[0037]

[0038] Assume that the stationary wavelet transform decomposes the electrical quantity M0 into i layers, which can obtain i+1 layers of sub-bands cd. The length of each sub-band is N, so all sub-bands can form a time-frequency matrix of size (i+1)×N. The electrical quantity after stationary wavelet decomposition can be expressed as:

[0039]

[0040] For each sub-band of the electrical quantity M0, the length is divided into equal parts according to the power frequency period (50 Hz). The division of the sub-band cdn of the nth layer can be expressed as:

[0041] cdn=[cdn(1);...;cdn(q);...;cdn(p)]q∈[1,p]

[0042] Where: p is the total number of time periods divided, and p = N / N T ; N T is the number of sampling points in each cycle, and N T =T×F s ; T is the power frequency period, F sis the sampling frequency. The feature extraction is performed on the q time period of the n-th layer sub-band cdn. The extracted feature quantities are as follows:

[0043] (1) Mean:

[0044] The average value is used to characterize the central tendency of the data and can describe the stability of the signal. Its calculation formula is:

[0045]

[0046] Among them, Mean j,q represents the average value of the jth electrical quantity in the qth time period; Indicates that in the nth layer sub-band, the length of the first time period is N T The sampling data sequence.

[0047] (2) Variance (Std):

[0048] Variance is used to measure the degree of dispersion of data, and its calculation formula is:

[0049]

[0050] Among them, Std j,q It represents the variance of the jth electrical quantity in the qth time period.

[0051] (3) Energy value (En):

[0052] The energy value is used to characterize the energy distribution of the data, and its calculation formula is:

[0053]

[0054] Among them, En j,q Indicates the energy value of the jth electrical quantity in the qth time period.

[0055] (4) Energy entropy (EEnt):

[0056] Energy entropy is used to characterize the degree of disorder in the energy sequence of data. The severity of disorder is positively correlated with the entropy value. Its calculation formula is:

[0057]

[0058] Where: En is the total energy value of the sub-band cdn; EEnt j,q It represents the energy entropy of the jth electrical quantity in the qth time period.

[0059] (5) Frequency entropy (FEnt):

[0060] The frequency entropy value is used to characterize the complexity and disorder of each sub-band, and its calculation formula is:

[0061]

[0062] Among them, FEnt j,q It represents the frequency entropy value of the jth electrical quantity in the qth time period.

[0063] The normalization process in the formula can be expressed as

[0064]

[0065] (6) Entropy weight (WEnt):

[0066] The entropy weight is used to characterize the proportion of the frequency entropy value of the nth layer q time period to the total frequency entropy value. This value can reflect the important characteristics of the data in a certain frequency band. Its calculation formula is:

[0067]

[0068] Among them, WEnt j,q It represents the entropy weight of the jth electrical quantity in the qth time period.

[0069] 4. Based on the feature extraction results and the constructed complete data set, a machine learning algorithm is used to perform weight analysis on the extracted features, calculate the contribution and correlation of each feature, determine the importance of the features, and sort them. After discarding features with lower weights, the remaining features are weighted and scored according to the weights. The feature weights are converted into specific scoring indicators, and a scoring criterion is constructed that associates feature values ​​with the degree of aging.

[0070] The specific process uses the Random Forest method to calculate weights and classify based on features. Random Forest is an ensemble learning algorithm that constructs multiple decision trees for classification and regression tasks. The basic idea of ​​Random Forest is to randomly select subsets of features and samples, construct multiple decision trees, and then combine the results of these decision trees through voting or averaging. The average score is also output along with the results for researchers to observe.

[0071] 5. Deploy sensors to collect multi-source data in real time and monitor the cable system. Ensure that sensors cover all critical areas and effectively capture various status information of the cable system. Build a sensor network, connect each sensor to the data acquisition system, and use wired or wireless communication technologies such as Ethernet and LoRa to achieve real-time data transmission and centralized storage. Develop a real-time monitoring platform to analyze and visualize the collected multi-source data in real time, providing real-time feedback on the operating status and health of the cable system.

[0072] 6. Set alarm thresholds based on existing aging data sets, issue early warnings for cable sections whose aging scores exceed the thresholds, and assess the remaining lifespan. By embedding the thresholds into the real-time monitoring system, the system automatically generates an aging score for each cable section's real-time data and compares it against the set thresholds.

[0073] 7. Synchronize the early warning information to the maintenance management platform to generate a detailed early warning report, including the specific location of the cable section, aging score, parameters exceeding the threshold, etc., to facilitate maintenance personnel to quickly locate and deal with problem areas.

[0074] 8. Check the alarmed section to evaluate the alarm accuracy. If the alarm is unreasonable, promptly revise the weight coefficient and scoring criteria.

Claims

1. A method for evaluating the aging state of cable insulation, characterized by: By fusing data collected by multiple sensors and combining it with intelligent algorithms, the aging status of the operating cable system can be evaluated. The following steps are included: Step 1: Collect operating data of cables of different ages, correlate these actual operating data with the age of the cables, environmental conditions, and load conditions, and construct a preliminary data set; Step 2: Generate simulation data to supplement the actual collected data and build a complete data set; Step 3: Extract features from the signals in the complete data set to extract characteristic indicators that are valuable for cable aging and performance evaluation; Step 4: Perform weight analysis on the extracted features; Step 5: Deploy sensors to collect multi-source data in real time and monitor the cable system; Step 6: Set an alarm threshold, issue an early warning for cable sections whose aging scores exceed the threshold, and evaluate the remaining lifespan; Step 7: Synchronize the warning information to the maintenance management platform and generate a detailed warning report; Step 8: Check the alarmed section to evaluate the alarm accuracy. If the alarm is unreasonable, promptly revise the weight coefficient and scoring criteria.

2. The method for evaluating cable insulation aging status according to claim 1, wherein: In step 2, the data set is supplemented by simulation technology. Simulation technology is used to simulate various possible cable aging scenarios, and their performance under different service years, operating conditions and fault conditions, generating a large amount of detailed simulation data. These simulation data, as a supplement, can effectively make up for the lack of field data and make the data set more comprehensive and diversified.

3. The method for evaluating cable insulation aging status according to claim 1, wherein: The specific process of step 3 is as follows: preprocess the collected sensor data, including data cleaning, denoising, and normalization, to ensure data quality and consistency; conduct in-depth data mining through time domain analysis, frequency domain analysis, and time-frequency analysis methods to extract key features.

4. The method for evaluating cable insulation aging status according to claim 3, wherein: The key features include the average value used to characterize the central tendency of the data, the variance used to measure the degree of dispersion of the data, the energy value used to characterize the energy distribution of the data, the energy entropy used to characterize the degree of disorder of the energy sequence of the data, the frequency entropy value used to characterize the complexity and disorder of each layer of sub-frequency band, and the entropy weight used to characterize the frequency entropy value of each layer and each time period as a percentage of the total frequency entropy value.

5. The method for evaluating cable insulation aging status according to claim 1, wherein: The specific process of step 4 is as follows: based on the feature extraction results and the constructed complete data set, a machine learning algorithm is used to perform weight analysis on the extracted features, calculate the contribution and correlation of each feature, determine the importance of the features, and sort them. After discarding features with lower weights, the remaining features are weighted and scored according to the weights. The feature weights are converted into specific scoring indicators, and a scoring criterion is constructed that associates feature values ​​with the degree of aging.

6. The method for evaluating cable insulation aging status according to claim 1, wherein: In step 4, weights are calculated based on the correlation between multi-source data and aging. Specifically, operating data, environmental data, and historical aging data of the cable are obtained from multiple sources to ensure the comprehensiveness and diversity of the data. The correlation between these multi-source data is also explored, especially the correlation with the cable aging process. Based on these correlations and the random forest algorithm, a scoring criterion for the association between eigenvalues ​​and aging degree was constructed.

7. The method for evaluating cable insulation aging status according to claim 1, characterized in that: The specific process of step 5 is as follows: establish a sensor network to ensure that sensors can cover all key areas, connect each sensor to the data acquisition system, and use wired or wireless communication technology to achieve real-time transmission and centralized storage of data; perform real-time analysis and visualization of the collected multi-source data to provide real-time feedback on the operating status and health of the cable system.

8. The method for evaluating cable insulation aging status according to claim 1, wherein: The early warning report described in step 7 includes the specific location of the cable section, the aging score, and the parameters exceeding the threshold.

Images