Cable insulation multi-parameter dynamic monitoring and fault pre-diagnosis system and implementation method

Abstract

This invention relates to the field of cable insulation monitoring and fault diagnosis technology, specifically disclosing a multi-parameter dynamic monitoring and fault prediction system for cable insulation and its implementation method. The system includes: a data acquisition module that collects insulation resistance and load current parameters and performs spatiotemporal registration to output a load dataset; an interference analysis module that filters insulation analysis segments and determines latent damage accumulation through energy accumulation analysis; an attenuation analysis module that calculates the attenuation reversibility value through coupled analysis of impact intensity and recovery efficiency, determining whether an attenuation warning signal is triggered; and a cluster intervention module that, after triggering the warning, divides the globally weakened impact protection segments into spatial clusters, constructs a reversible prediction model by combining reversible parameters and impact parameters, and outputs the attenuation conversion time. The method is used to achieve multi-parameter dynamic monitoring and fault prediction of cable insulation. This invention improves the accuracy of prediction through multi-parameter coupling and global cluster analysis, providing a basis for cable operation and maintenance.

Description

Technical Field

[0001] This invention relates to the field of cable insulation monitoring and fault diagnosis technology, specifically to a system and method for dynamic monitoring and fault pre-diagnosis of multiple parameters of cable insulation. Background Technology

[0002] As the core transmission carrier of the power system, the insulation performance of cables determines the stability and security of power transmission. With the increase in power system capacity and the extension of operation cycle, cable insulation is susceptible to latent damage caused by load impact and environmental factors. If it cannot be identified and controlled in time, it can easily lead to power failure and affect power supply.

[0003] Existing technologies are insufficient in identifying the cumulative trend of latent damage to cables, and usually only trigger warnings when the damage has developed to a certain extent, thus missing the early intervention window. Existing technologies lack global cluster analysis for scattered high-risk monitoring segments, making it difficult to allocate resources in a targeted manner during operation and maintenance, which can easily lead to blind spots in management or over-maintenance. Overall, the efficiency of early warning and management needs to be improved.

[0004] Therefore, the present invention provides a dynamic monitoring and fault pre-diagnosis system for multiple parameters of cable insulation and a method for its implementation. Summary of the Invention

[0005] The purpose of this invention is to provide a system and method for dynamic monitoring and fault pre-diagnosis of multiple parameters of cable insulation, in order to solve the aforementioned background problems.

[0006] The objective of this invention can be achieved through the following technical solutions:

[0007] The cable insulation multi-parameter dynamic monitoring and fault pre-diagnosis system includes the following modules:

[0008] Data acquisition module: used to collect insulation parameters of cable monitoring sections and perform spatiotemporal registration analysis, outputting load dataset;

[0009] Interference analysis module: Based on the load dataset, the monitoring segments are classified to obtain monitoring segments with sudden rise and fall and micro-fluctuation features are extracted. Based on the micro-fluctuation features, insulation analysis segments are screened, and latent damage analysis is performed on the insulation analysis segments to determine whether there is latent damage accumulation in the insulation analysis segments.

[0010] Attenuation Analysis Module: If there is latent damage accumulation, the insulation analysis section is weakened based on the load current to obtain the weakened section. The attenuation reversibility is analyzed on the weakened section to obtain the attenuation reversibility value and determine whether the attenuation signal is triggered.

[0011] Cluster Intervention Module: If the attenuation warning signal is triggered, the global impact weakening segment is divided into multiple spatial clusters, a reversible prediction model is constructed, the reversible parameters and impact parameters of each spatial cluster are obtained and input into the model, the attenuation conversion time is output, and fault intervention is performed on the spatial clusters.

[0012] As a further aspect of the present invention, the spatiotemporal registration analysis is performed as follows:

[0013] The load current and insulation resistance values ​​of each monitoring segment are obtained as insulation parameters, and the spatial coordinate codes and sampling timestamps of the load current and insulation resistance values ​​are extracted.

[0014] According to the same spatial coordinate encoding, the insulation resistance value and load current value of the monitored segmented cable are matched according to the insulation resistance value and load current value with the same sampling timestamp to form a four-data set of time-space-insulation resistance value-load current value.

[0015] Obtain the four-data set of all monitoring segments at N sampling timestamps and construct the load dataset.

[0016] As a further aspect of the present invention, the method for extracting the micro-fluctuation features is as follows:

[0017] Obtain the preset steady-state range of the load current, and determine the recovery period of the recovery phase based on the steady-state range of the load current;

[0018] Extract the frequency and directional fluctuation ratio of micro-fluctuations within the recovery period, and use both as micro-fluctuation characteristics;

[0019] The directional volatility ratio includes both positive volatility ratio and negative volatility ratio.

[0020] As a further aspect of the present invention: the method for extracting the directional fluctuation ratio is as follows:

[0021] Obtain the insulation resistance reference value, and construct the fluctuation start and end judgment conditions and effective fluctuation conditions based on the insulation resistance reference value. When the insulation resistance value of the monitored segment meets the fluctuation start and end judgment conditions and effective fluctuation conditions, it is judged as an effective fluctuation.

[0022] If the insulation resistance value is lower than the insulation resistance reference value during the effective fluctuation within the recovery period, the effective fluctuation is marked as a negative deviation fluctuation; if it is higher than the insulation resistance reference value, it is marked as a positive deviation fluctuation.

[0023] The proportion of negative deviations in insulation resistance value during the recovery period is calculated as the negative fluctuation ratio.

[0024] The positive fluctuation ratio is calculated as the proportion of positive deviations in insulation resistance value within the recovery period relative to the total number of fluctuations.

[0025] As a further aspect of the present invention, the method for performing the latent damage analysis is as follows:

[0026] If the negative fluctuation ratio of a monitoring segment is higher than the positive fluctuation ratio for M consecutive monitoring cycles, the monitoring segment will be marked as an insulation analysis segment.

[0027] For each effective negative fluctuation in the insulation analysis section during the recovery period, the negative fluctuation energy of the recovery period is obtained through the negative fluctuation energy formula.

[0028] The total negative fluctuation energy of the recovery cycle is calculated to obtain the cumulative energy of the cycle.

[0029] The periodic energy of the insulation analysis segment over M monitoring cycles is obtained and linear regression is performed. If the slope of the fit is positive, the cumulative periodic energy of the insulation analysis segment over the M monitoring cycles shows an increasing trend, indicating the presence of latent damage accumulation.

[0030] As a further aspect of the present invention, the method for performing the weakening identification is as follows:

[0031] Calculate the absolute difference between the baseline value of insulation resistance and the extreme value of load current as the absolute change.

[0032] Obtain the absolute changes of the insulation analysis section over M monitoring periods and construct an absolute change sequence;

[0033] Cluster analysis is performed on the absolute change sequence, and the absolute change is divided into different intensity levels based on the cluster analysis results;

[0034] Calculate the average fluctuation frequency of each intensity level within the current monitoring period, and perform KL divergence calculation with the normal fluctuation benchmark value to obtain the deviation of each intensity level;

[0035] Based on the deviation, a weakening judgment criterion is constructed. If the deviation of each strength level of the insulation analysis segment meets the weakening judgment criterion, then the insulation analysis segment is a weakened segment.

[0036] As a further aspect of the present invention, the method for performing the attenuation reversibility analysis is as follows:

[0037] Obtain the recovery efficiency - change data set and the historical recovery efficiency - change data set;

[0038] A nonlinear regression algorithm was used to fit the recovery efficiency-change data set and the historical recovery efficiency-change data set of the weakened segment to obtain the current curve and the historical curve.

[0039] Feature extraction is performed on the two curves to obtain the recovery attenuation coefficient and the slope deterioration ratio;

[0040] The average value of the recovery attenuation coefficients is obtained by summing the attenuation coefficients of all intensity levels.

[0041] Construct a proportional equation, input the mean attenuation coefficient and the slope deterioration ratio into the proportional equation, and obtain the attenuation reversibility value.

[0042] As a further aspect of the present invention, the feature extraction process is performed as follows:

[0043] Extract the characteristic impact intensity points of the two curves at each intensity level;

[0044] Calculate the percentage difference between the recovery efficiency and the historical recovery efficiency at the characteristic impact intensity point of each intensity level for the current curve and the historical curve, and obtain the recovery attenuation coefficient for each intensity level.

[0045] Calculate the average slope of the current curve from medium to high strength levels, and use it as the current strength slope;

[0046] Calculate the average slope of the historical curve from medium to high strength levels, and use it as the historical strength slope.

[0047] Calculate the absolute ratio of the current intensity slope to the historical intensity slope, and use this as the slope deterioration ratio.

[0048] As a further aspect of the present invention: the method for outputting the attenuation conversion time is as follows:

[0049] For each spatial cluster, calculate the total number of defensive weakening segments within each cluster, i.e., the total number of segments in the cluster;

[0050] The number of irreversible decay segments in each spatial cluster is obtained, and the ratio of this number to the total number of segments in the cluster is calculated to obtain the cluster non-inverse ratio.

[0051] The cluster inverse ratio and decay inverse value of the spatial cluster are used as inverse parameters.

[0052] The absolute change in load impact and the negative fluctuation ratio of each weakened section in the spatial cluster are used as impact parameters.

[0053] A reversible prediction model is constructed based on reversible and impact parameters, and the decay transformation time of all reversibly decayed impact-weakening segments within the spatial cluster into irreversible impact-weakening segments is output.

[0054] The beneficial effects of this invention are:

[0055] (1) Based on the load dataset, the load type classification of the monitoring segment is realized, the micro-fluctuation characteristics of the sudden rise and fall segment are extracted in a targeted manner, the insulation analysis segment is screened by the negative fluctuation ratio, and the latent damage accumulation is determined by the negative fluctuation energy accumulation trend. This can identify the potential damage accumulation risk when the cable insulation does not show obvious faults, which is conducive to the early identification of high-risk monitoring segments.

[0056] (2) For the segmented identification of the impact-weakened segment with latent damage accumulation, the current and historical recovery efficiency-change data sets are constructed and fitted curves are fitted. The recovery attenuation coefficient and slope deterioration ratio of the characteristic impact intensity points are combined to calculate the attenuation reversible value. The impact resistance attenuation property of the impact-weakened segment is effectively analyzed, which provides a basis for whether to start the fault intervention process. This is conducive to reducing the over-processing of reversible fluctuations or the under-processing of irreversible attenuation.

[0057] (3) After triggering the attenuation warning, the global impact weakening segment is divided into spatial clusters. The reversible prediction model is constructed by combining the cluster non-reversible ratio (reversible parameter) with the absolute change of load impact and the negative fluctuation ratio (impact parameter). The attenuation conversion time is output and compared with the threshold to determine intervention or monitoring. This is conducive to the control of cable insulation attenuation risk, realizing differentiated treatment of different spatial clusters, prioritizing response to high-risk clusters, and improving the systematicness and pertinence of cable insulation risk control. Attached Figure Description

[0058] The invention will now be further described with reference to the accompanying drawings.

[0059] Figure 1 This is a block diagram of the cable insulation multi-parameter dynamic monitoring and fault pre-diagnosis system of the present invention;

[0060] Figure 2 This is a flowchart of the recovery feature extraction process in this invention.

[0061] Figure 3 This is a flowchart of the method for dynamic monitoring and fault pre-diagnosis of multiple parameters of cable insulation in this invention. Detailed Implementation

[0062] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0063] Example 1

[0064] Please see Figure 1As shown, this invention is a dynamic monitoring and fault pre-diagnosis system for multiple parameters of cable insulation, comprising the following modules:

[0065] Data acquisition module: used to collect insulation parameters of cable monitoring sections and perform spatiotemporal registration analysis, outputting load dataset;

[0066] The method for collecting cable insulation parameters and performing spatiotemporal registration analysis is as follows:

[0067] Preferably, the cable is divided into monitoring segments at fixed intervals, and a spatial coordinate code is assigned to each monitoring segment; an insulation resistance (IR) sensor is deployed on the cable of each monitoring segment to output the insulation resistance value of the sampling timestamp within the monitoring period;

[0068] Simultaneously, the load current value of each monitoring segment is acquired, and the load current value and insulation resistance value of each insulation segment are timestamped and synchronously calibrated. The processed insulation resistance value and load current value are then bound to the spatial coordinates of the monitoring segment to generate insulation resistance value and load current value with spatial coordinates.

[0069] It should be noted that insulation resistance and load current are used as the insulation parameters of the cable; current sensors are deployed on the cable in each monitoring segment to synchronously output the load current value corresponding to the sampling timestamp within the monitoring period, and at the same time, the load current value of each monitoring segment is acquired.

[0070] According to the same spatial coordinate encoding, the insulation resistance value and load current value of the monitored segmented cable are matched according to the insulation resistance value and load current value with the same sampling timestamp to form a four-data set of time-space-insulation resistance value-load current value.

[0071] Obtain the four-data set of all monitoring segments at N sampling timestamps to construct the load dataset;

[0072] Interference analysis module: Based on the load dataset, the monitoring segments are classified to obtain monitoring segments with sudden rise and fall and micro-fluctuation features are extracted. Based on the micro-fluctuation features, insulation analysis segments are screened, and latent damage analysis is performed on the insulation analysis segments to determine whether there is latent damage accumulation in the insulation analysis segments.

[0073] The method for classifying monitoring segments based on the load dataset to obtain sudden rise and fall type monitoring segments and extracting micro-fluctuation features is as follows:

[0074] The load current sequence is constructed by extracting the load current values ​​of each monitoring segment at N sampling timestamps from the load dataset.

[0075] Preferably, N=50;

[0076] A load classification model is constructed using a density clustering algorithm. The load current sequence is input into the load classification model, which then outputs the type of load current change.

[0077] Those skilled in the art will understand that the method of constructing a load classification model using density clustering algorithm is as follows: First, construct the load current sequence corresponding to each monitoring segment; then, construct the load classification model based on the density clustering algorithm (DBSCAN algorithm). Based on the conventional fluctuation characteristics of cable load current, set the neighborhood radius and the minimum number of samples for core point determination; then, convert the fluctuation characteristics of the load current sequence (such as the amplitude, duration, and frequency of current surges or drops) into feature vectors recognizable by the model; label load current sequence samples of three change types—sudden surge / drop, stepped fluctuation, and random fluctuation—to train and optimize the model, determining the clusters corresponding to different change types; finally, input the load current sequence of the monitoring segment to be classified into the trained load classification model. The load classification model calculates the density correlation between the sequence and each cluster, outputting the corresponding load current change type.

[0078] The variation types include: sudden rise and fall type, stepped fluctuation type, and random fluctuation type. The labeling rules for the samples are as follows: sudden rise and fall type: the amplitude of a single current fluctuation exceeds 30% of the rated load and the duration is <10 seconds; stepped fluctuation type: the current changes in steps of 5%-10% of the rated load, and the stable duration between steps is ≥30 seconds; random fluctuation type: the current fluctuates irregularly within ±5% of the rated load, and the fluctuation frequency is >5 times / minute. The sample source is the measured current sequence of similar cables under normal operation.

[0079] Obtain the preset steady-state range of the load current and the characteristic markers of the steady-state range. If the load current of the sudden rise and fall monitoring segment meets the characteristic markers and is no longer in a sudden rise and fall state, then perform recovery feature extraction to obtain micro-fluctuation features.

[0080] It should be noted that, taking into account the rated load range of the cable and the load current fluctuation range during historical normal operation, a range in which the load current is kept stable (such as the range of ±10% of the rated load current) is preset as the steady-state range.

[0081] For example, restoring the load current to a stable range of ±5% before the sudden rise or fall is used as a characteristic indicator of the steady-state range;

[0082] like Figure 2 As shown, the method for performing feature extraction is as follows:

[0083] S201. Determine the recovery period of the recovery phase based on the steady-state range of the load current;

[0084] Preferably, the moment when the sudden rise and fall ends (i.e. the moment when the current begins to fall from the peak or valley) is taken as the start time of the recovery cycle, and the moment when the load current is kept within ±5% of the load current value before the impact for 5 consecutive minutes is taken as the end time.

[0085] The time range of the start and end times is used as the recovery period;

[0086] S202. Extract the frequency and directional fluctuation ratio of micro-fluctuations within the recovery period, and use both as micro-fluctuation features;

[0087] The average insulation resistance value of the monitored segment before the sudden rise or fall occurs and when the load current is within a stable range is obtained, and this value is used as the reference value for insulation resistance.

[0088] Based on the insulation resistance reference value, the fluctuation start and end judgment conditions and the effective fluctuation conditions are constructed. When the insulation resistance value of the monitored segment meets the fluctuation start and end judgment conditions and the effective fluctuation conditions, it is judged as an effective fluctuation.

[0089] For example, the method for constructing the fluctuation start and end judgment conditions and the effective fluctuation conditions is as follows: the average insulation resistance of the monitoring segment before the sudden rise and fall occurs and when the load current is in the stable range is used as the insulation resistance reference value. On this basis, the fluctuation start and end judgment conditions are constructed. When the insulation resistance value of the monitoring segment deviates from the insulation resistance reference value by ±5%, it is determined that the fluctuation has started. When the insulation resistance value returns to the insulation resistance reference value within ±5% and lasts for more than 30 seconds, it is determined that the fluctuation has ended. At the same time, the effective fluctuation conditions are constructed. The insulation resistance value must deviate from the reference value by more than 5%, and the duration of this deviation must be greater than 10 seconds to exclude short-term fluctuations caused by sensor instantaneous errors.

[0090] The recovery period is divided into multiple sub-periods according to a fixed time window. The occurrence rate of effective fluctuations in each sub-period is obtained, and the average occurrence rate of effective fluctuations in all sub-periods is calculated as the micro-fluctuation frequency.

[0091] Preferably, the number of sub-period divisions is 10;

[0092] The directional volatility ratio includes: positive volatility ratio and negative volatility ratio;

[0093] Preferably, the method for obtaining the positive volatility ratio and the negative volatility ratio is as follows:

[0094] If the insulation resistance value is lower than the insulation resistance reference value during the effective fluctuation within the recovery period, the effective fluctuation will be marked as a negative deviation fluctuation.

[0095] If the insulation resistance value is higher than the insulation resistance reference value during the effective fluctuation within the recovery period, it is marked as a positive deviation fluctuation.

[0096] If the insulation resistance value during the effective fluctuation within the recovery period is equal to the insulation resistance reference value, no action is taken.

[0097] The proportion of negative deviations in insulation resistance value during the recovery period is calculated as the negative fluctuation ratio.

[0098] The proportion of positive deviations in insulation resistance value within the recovery period to the total number of fluctuations is calculated as the positive fluctuation ratio.

[0099] Based on the micro-fluctuation characteristics, insulation analysis segments are screened, and energy accumulation analysis is performed on the insulation analysis segments to determine whether there is latent damage accumulation in the insulation analysis segments.

[0100] The method for screening insulation analysis segments based on micro-fluctuation characteristics, performing latent damage analysis on the insulation analysis segments, and determining whether there is latent damage accumulation in the insulation analysis segments is as follows:

[0101] If the negative fluctuation ratio of a monitoring segment is higher than the positive fluctuation ratio for M consecutive monitoring cycles, the monitoring segment will be marked as an insulation analysis segment.

[0102] Preferably, M=10;

[0103] The method for energy accumulation analysis of the insulation analysis section is as follows: For each effective negative fluctuation of the insulation analysis section within the recovery period, the negative fluctuation energy formula is used: Obtain the negative fluctuation energy of the i-th recovery cycle. ;

[0104] in, This is the reference value for insulation resistance. Let be the minimum insulation resistance value during the i-th fluctuation. The duration of the i-th fluctuation;

[0105] The total negative fluctuation energy of the recovery cycle is calculated to obtain the cumulative energy of the cycle.

[0106] The periodic energy of the insulation analysis segment over M monitoring cycles is obtained and linear regression is performed. If the slope of the fit is positive, the cumulative periodic energy of the insulation analysis segment over M monitoring cycles shows an increasing trend, indicating the presence of latent damage accumulation.

[0107] If the value is negative, the cumulative energy of the insulation analysis section over M monitoring cycles shows a decreasing trend.

[0108] Example 2

[0109] Please see Figure 1 As shown, this invention is a dynamic monitoring and fault prediction system for multiple parameters of cable insulation, and also includes the following modules:

[0110] Attenuation Analysis Module: If there is latent damage accumulation, the insulation analysis section is weakened based on the load current to obtain the weakened section. The attenuation reversibility is analyzed on the weakened section to obtain the attenuation reversibility value and determine whether the attenuation signal is triggered.

[0111] Among them, the method for identifying weakened insulation analysis sections based on load current to obtain the weakened insulation sections is as follows:

[0112] Preferably, if there is latent damage accumulation in the insulation analysis section, the extreme value of the load current during the sudden rise and fall is obtained;

[0113] Calculate the absolute difference between the baseline value of insulation resistance and the extreme value of load current as the absolute change.

[0114] Obtain the absolute changes of the insulation analysis section over M monitoring periods and construct an absolute change sequence;

[0115] Cluster analysis is performed on the absolute change sequence, and the absolute change is divided into different intensity levels based on the cluster analysis results;

[0116] It should be noted that the cluster analysis is performed as follows: based on the absolute change sequence of the insulation analysis section within M monitoring cycles (the set of absolute differences between the insulation resistance reference value and the load current extreme value), a clustering algorithm (K-means clustering) is used to group the absolute differences in the sequence according to their magnitude, fluctuation frequency, and other characteristics. Then, based on the differences in the absolute difference range and cluster density of different groups, low, medium, and high intensity levels are distinguished.

[0117] Obtain historical data of insulation analysis sections under no-damage accumulation, and obtain the average value of the fluctuation frequency of the corresponding strength level as the normal fluctuation benchmark value under no-damage accumulation.

[0118] Among them, the historical data without damage accumulation, that is, the data such as the absolute change and fluctuation frequency in the historical monitoring period in which no latent damage accumulation was determined, are integrated to obtain the historical data without damage accumulation.

[0119] Calculate the average fluctuation frequency of each intensity level within the current monitoring period, and perform KL divergence calculation with the normal fluctuation benchmark value to obtain the deviation of each intensity level;

[0120] Those skilled in the art will understand that a normal distribution is constructed using the mean μ and standard deviation σ of the cumulative historical data of non-damage for each defined intensity level. ,in The mean fluctuation frequency is the average of the last 100 damage-free cycles, and σ is the corresponding standard deviation; calculate the normal distribution of the mean fluctuation frequency of the current cycle. Using the KL divergence formula Get , For historical distribution probability, Given the current probability distribution, assign each intensity level... As the deviation of each intensity level;

[0121] Based on the deviation, a weakening judgment criterion is constructed. If the deviation of each strength level of the insulation analysis segment meets the weakening judgment criterion, then the insulation analysis segment is a weakened segment.

[0122] For example, the weakening judgment criterion is constructed as follows: taking the deviation calculated by KL divergence of each strength level (low, medium, high) as the core, the thresholds are set as follows: the deviation of low strength level ≥ 0.3, the deviation of medium strength level ≥ 0.5, and the deviation of high strength level ≥ 0.7. If the deviation of the insulation analysis section of the low, medium and high strength levels reaches or exceeds the corresponding set thresholds in the current monitoring period, it is determined that it meets the weakening judgment criterion.

[0123] The method for performing attenuation reversibility analysis on the weakening segment to obtain the attenuation reversibility value and determine whether the attenuation signal is triggered is as follows:

[0124] Calculate the reciprocal of the recovery period of the weakened phase of the defense, and use it as the recovery efficiency;

[0125] Obtain the recovery efficiency and absolute change of the weakened segment in each monitoring period, and integrate them according to the sampling timestamp dimension to construct the recovery efficiency-change data group of the weakened segment.

[0126] At the same time, the absolute change of the weakened segment under historical no-damage accumulation is obtained, and the corresponding recovery efficiency is obtained. The recovery efficiency and the absolute change under no-damage accumulation are integrated to construct a historical recovery efficiency-change data set.

[0127] A nonlinear regression algorithm (polynomial regression algorithm, order 2) is used to fit the recovery efficiency-change data set and the historical recovery efficiency-change data set of the weakened segment to obtain the current curve and the historical curve.

[0128] Perform reversible decay analysis on the two curves to determine whether the weakening segment of the defense is in an irreversible decay state.

[0129] The process of performing reversible decay analysis is as follows:

[0130] The average absolute change in each intensity level range (low, medium, and high) is used as the characteristic impact intensity value for each intensity level.

[0131] For the current curve and the historical curve, select the absolute change with the same value according to the characteristic impact intensity value as the characteristic impact intensity point for each intensity level;

[0132] Calculate the percentage difference between the recovery efficiency and the historical recovery efficiency at the characteristic impact intensity point of each intensity level for the current curve and the historical curve, and obtain the recovery attenuation coefficient for each intensity level.

[0133] Calculate the average slope of the current curve from medium to high strength levels, and use it as the current strength slope;

[0134] Calculate the average slope of the historical curve from medium to high strength levels, and use it as the historical strength slope.

[0135] It should be noted that if the current curve and the historical curve have multiple medium to high intensity levels, the average slope of all level ranges is calculated as the slope of the corresponding intensity.

[0136] Calculate the absolute ratio of the current intensity slope to the historical intensity slope, and use it as the slope deterioration ratio C2;

[0137] The average value of the recovery attenuation coefficients of all intensity levels is obtained by summing and averaging the values.

[0138] By constructing a proportional equation: The attenuation reversibility value is obtained;

[0139] It is understandable that obtaining the decay reversibility value serves the following purpose:

[0140] Function 1: Used to determine the attenuation of the impact resistance of the weakened section. The attenuation reversible value is compared with the preset attenuation critical threshold. If the attenuation reversible value is lower than or equal to the threshold, the impact resistance is determined to be irreversibly attenuated and an attenuation warning signal is triggered. If it is higher than the threshold, it is determined to be reversible fluctuation.

[0141] Function 2: Provides key data for building a reversible prediction model for the cluster intervention module, helping the model output the decay transformation time of the reversible decay of the weakening segment within the spatial cluster into the irreversible segment, providing a basis for subsequent fault intervention or continuous monitoring.

[0142] The attenuation reversible value is compared with the preset attenuation critical threshold. If the attenuation critical value is lower than or equal to the preset attenuation critical threshold, it is determined that the impact resistance is irreversibly attenuated, and an attenuation warning signal is triggered.

[0143] If the attenuation threshold is higher than the preset attenuation threshold, then it is determined to be a reversible fluctuation.

[0144] It should be noted that the attenuation critical threshold can be pre-determined by professionals in the field through historical data accumulated without damage (such as the average attenuation coefficient and slope deterioration ratio of the weakened section under historical undamaged conditions) and the design impact resistance of the cable insulation. This threshold is then compared with the calculated attenuation reversibility value to determine whether the attenuation is irreversible.

[0145] Cluster Intervention Module: If the attenuation warning signal is triggered, the global impact weakening segment is divided into multiple spatial clusters, a reversible prediction model is constructed, the reversible parameters and impact parameters of each spatial cluster are obtained and input into the model, the attenuation conversion time is output, and fault intervention is performed on the spatial clusters.

[0146] The method for dividing the global defense weakening segment into multiple spatial clusters is as follows:

[0147] Preferably, the spatial coordinates, attenuation reversible values, and irreversible attenuation determination results of all weakened segments are obtained;

[0148] It should be noted that the initial remaining lifetime value is set by those skilled in the art;

[0149] Based on the spatial coordinates of the weakened segments, all weakened segments are clustered to obtain multiple spatial clusters.

[0150] For example, the cluster partitioning method is as follows: the neighborhood radius is ≤500 meters, and the core point is determined by the condition that it contains at least 3 defensive weakening segments;

[0151] Segments that meet the neighborhood condition are grouped into the same spatial cluster, while isolated segments that do not meet the condition are set up as separate single-segment clusters.

[0152] Multiple spatial clusters are obtained (such as cluster 1, cluster 2, ..., cluster Z), and each cluster is labeled with the number of the weakened segment and its spatial range.

[0153] It should be noted that Z is the maximum value of the cluster number;

[0154] The method for constructing a reversible prediction model, obtaining the reversible and impact parameters of each spatial cluster and inputting them into the model, outputting the decay conversion time, and intervening in the fault of the spatial cluster is as follows:

[0155] For each spatial cluster, calculate the total number of defensive weakening segments within each cluster, i.e., the total number of segments in the cluster;

[0156] The number of irreversible decay segments in each spatial cluster is obtained, and the ratio of this number to the total number of segments in the cluster is calculated to obtain the cluster non-inverse ratio.

[0157] The cluster inverse ratio and decay inverse value of the spatial cluster are used as inverse parameters.

[0158] The absolute change in load impact and the negative fluctuation ratio of each weakened section in the spatial cluster are used as impact parameters.

[0159] A reversible prediction model is constructed based on reversible parameters and impact parameters, and the decay conversion time of all reversible decaying shock-weakening segments in the spatial cluster into irreversible shock-weakening segments is output.

[0160] The decay conversion time is compared with the preset conversion control threshold. If the decay conversion time is lower than or equal to the preset conversion control threshold, a fault intervention signal is triggered, the spatial coordinate range and decay conversion time of the spatial cluster are recorded, and the signal is sent to the control system.

[0161] Understandably, the method for constructing the reversible transformation prediction model is as follows: For each spatial cluster, based on sample data of reversible attenuation and weakening segments transforming into irreversible segments in historical monitoring, the aforementioned reversible parameters and impact parameters are used as input features, with the actual transformation time as the output target. Exponential regression is used as the nonlinear regression form, the loss function is the mean squared error (MSE), and the iteration termination condition is MSE < 0.01. The model training samples need to include 50 sets of reversible → irreversible transformation cases. Each set of samples covers the cluster inverse ratio, attenuation reversibility value, average absolute change, average negative fluctuation ratio, and actual transformation time to realize the construction of the reversible transformation prediction model.

[0162] The conversion control threshold is set by professionals in the field by combining the rated operating parameters of the cable, the insulation life design standard, the actual operation and maintenance response cycle, and the statistical data of the attenuation conversion time of the reversible attenuation and weakening section in the same spatial cluster in historical monitoring. The conversion control threshold is set in a comprehensive manner to ensure that the threshold can reserve sufficient operation and maintenance intervention time while reducing the risk of irreversible attenuation due to excessively high threshold.

[0163] If the decay conversion time exceeds the preset conversion control threshold, the decay conversion time will be continuously monitored.

[0164] Example 3

[0165] like Figure 3 As shown, the method for dynamic monitoring and fault pre-diagnosis of multiple parameters of cable insulation includes the following steps:

[0166] Step 1: Collect the insulation parameters of the cable and perform spatiotemporal registration analysis to output the load dataset;

[0167] Step 2: Classify the monitoring segments based on the load dataset to obtain monitoring segments with sudden rises and falls and extract micro-fluctuation features. Based on the micro-fluctuation features, screen the insulation analysis segments and perform latent damage analysis on the insulation analysis segments to determine whether there is latent damage accumulation in the insulation analysis segments.

[0168] Step 3: If there is latent damage accumulation, the insulation analysis section is weakened based on the load current to obtain the weakened section. The attenuation reversibility of the weakened section is analyzed to obtain the attenuation reversibility value and determine whether the attenuation signal is triggered.

[0169] Step 4: If the attenuation warning signal is triggered, the global impact weakening segment is divided into multiple spatial clusters, a reversible prediction model is constructed, the reversible parameters and impact parameters of each spatial cluster are obtained and input into the model, the attenuation conversion time is output, and fault intervention is performed on the spatial clusters.

[0170] The foregoing has provided a detailed description of one embodiment of the present invention, but this description is merely a preferred embodiment and should not be construed as limiting the scope of the invention. All equivalent variations and modifications made within the scope of the present invention should still fall within the scope of the present invention.

Claims

1. A dynamic monitoring and fault prediction system for multiple parameters of cable insulation, characterized in that: Includes the following modules: Data acquisition module: used to collect insulation parameters of cable monitoring sections and perform spatiotemporal registration analysis, outputting load dataset; Interference analysis module: Based on the load dataset, the monitoring segments are classified to obtain monitoring segments with sudden rise and fall and micro-fluctuation features are extracted. Based on the micro-fluctuation features, insulation analysis segments are screened, and latent damage analysis is performed on the insulation analysis segments to determine whether there is latent damage accumulation in the insulation analysis segments. Attenuation Analysis Module: If there is latent damage accumulation, the insulation analysis section is weakened based on the load current to obtain the weakened section. The attenuation reversibility is analyzed on the weakened section to obtain the attenuation reversibility value and determine whether to trigger the attenuation warning signal. Cluster Intervention Module: If the attenuation warning signal is triggered, the global impact weakening segment is divided into multiple spatial clusters, a reversible prediction model is constructed, the reversible parameters and impact parameters of each spatial cluster are obtained and input into the model, the attenuation conversion time is output, and fault intervention is performed on the spatial clusters.

2. The cable insulation multi-parameter dynamic monitoring and fault pre-diagnosis system according to claim 1, characterized in that: The spatiotemporal registration analysis is performed as follows: The load current and insulation resistance values ​​of each monitoring segment are obtained as insulation parameters, and the spatial coordinate codes and sampling timestamps of the load current and insulation resistance values ​​are extracted. According to the same spatial coordinate encoding, the insulation resistance value and load current value of the monitored segmented cable are matched according to the insulation resistance value and load current value with the same sampling timestamp to form a four-data set of time-space-insulation resistance value-load current value. Obtain the four-data set of all monitoring segments at N sampling timestamps and construct the load dataset.

3. The cable insulation multi-parameter dynamic monitoring and fault pre-diagnosis system according to claim 1, characterized in that: The method for extracting the micro-fluctuation features is as follows: Obtain the preset steady-state range of the load current, and determine the recovery period of the recovery phase based on the steady-state range of the load current; Extract the frequency and directional fluctuation ratio of micro-fluctuations within the recovery period, and use both as micro-fluctuation characteristics; The directional volatility ratio includes both positive volatility ratio and negative volatility ratio.

4. The cable insulation multi-parameter dynamic monitoring and fault pre-diagnosis system according to claim 3, characterized in that: The method for extracting the directional fluctuation ratio is as follows: Obtain the insulation resistance reference value, and construct the fluctuation start and end judgment conditions and effective fluctuation conditions based on the insulation resistance reference value. When the insulation resistance value of the monitored segment meets the fluctuation start and end judgment conditions and effective fluctuation conditions, it is judged as an effective fluctuation. If the insulation resistance value is lower than the insulation resistance reference value during the effective fluctuation within the recovery period, the effective fluctuation is marked as a negative deviation fluctuation; if it is higher than the insulation resistance reference value, it is marked as a positive deviation fluctuation. The proportion of negative deviations in insulation resistance value during the recovery period is calculated as the negative fluctuation ratio. The positive fluctuation ratio is calculated as the proportion of positive deviations in insulation resistance value within the recovery period relative to the total number of fluctuations.

5. The cable insulation multi-parameter dynamic monitoring and fault pre-diagnosis system according to claim 1, characterized in that: The method for performing the latent damage analysis is as follows: If the negative fluctuation ratio of a monitoring segment is higher than the positive fluctuation ratio for M consecutive monitoring cycles, the monitoring segment will be marked as an insulation analysis segment. For each effective negative fluctuation in the insulation analysis section during the recovery period, the negative fluctuation energy of the recovery period is obtained through the negative fluctuation energy formula. The total negative fluctuation energy of the recovery cycle is calculated to obtain the cumulative energy of the cycle. The periodic energy of the insulation analysis segment over M monitoring cycles is obtained and linear regression is performed. If the slope of the fit is positive, the cumulative periodic energy of the insulation analysis segment over the M monitoring cycles shows an increasing trend, indicating the presence of latent damage accumulation.

6. The cable insulation multi-parameter dynamic monitoring and fault pre-diagnosis system according to claim 1, characterized in that: The weakening identification is performed as follows: Calculate the absolute difference between the baseline value of insulation resistance and the extreme value of load current as the absolute change. Obtain the absolute changes of the insulation analysis section over M monitoring periods and construct an absolute change sequence; Cluster analysis is performed on the absolute change sequence, and the absolute change is divided into different intensity levels based on the cluster analysis results; Calculate the average fluctuation frequency of each intensity level within the current monitoring period, and perform KL divergence calculation with the normal fluctuation benchmark value to obtain the deviation of each intensity level; Based on the deviation, a weakening judgment criterion is constructed. If the deviation of each strength level of the insulation analysis segment meets the weakening judgment criterion, then the insulation analysis segment is a weakened segment.

7. The cable insulation multi-parameter dynamic monitoring and fault pre-diagnosis system according to claim 1, characterized in that: The method for performing the attenuation reversibility analysis is as follows: Obtain the recovery efficiency - change data set and the historical recovery efficiency - change data set; A nonlinear regression algorithm was used to fit the recovery efficiency-change data set and the historical recovery efficiency-change data set of the weakened segment to obtain the current curve and the historical curve. Feature extraction is performed on the two curves to obtain the recovery attenuation coefficient and the slope deterioration ratio; The average value of the recovery attenuation coefficients is obtained by summing the attenuation coefficients of all intensity levels. Construct a proportional equation, input the mean attenuation coefficient and the slope deterioration ratio into the proportional equation, and obtain the attenuation reversibility value.

8. The cable insulation multi-parameter dynamic monitoring and fault pre-diagnosis system according to claim 7, characterized in that: The feature extraction process is performed as follows: Extract the characteristic impact intensity points of the two curves at each intensity level; Calculate the percentage difference between the recovery efficiency and the historical recovery efficiency at the characteristic impact intensity point of each intensity level for the current curve and the historical curve, and obtain the recovery attenuation coefficient for each intensity level. Calculate the average slope of the current curve from medium to high strength levels, and use it as the current strength slope; Calculate the average slope of the historical curve from medium to high strength levels, and use it as the historical strength slope. Calculate the absolute ratio of the current intensity slope to the historical intensity slope, and use this as the slope deterioration ratio.

9. The cable insulation multi-parameter dynamic monitoring and fault pre-diagnosis system according to claim 1, characterized in that: The method for outputting the decay conversion time is as follows: For each spatial cluster, calculate the total number of defensive weakening segments within each cluster, i.e., the total number of segments in the cluster; The number of irreversible decay segments in each spatial cluster is obtained, and the ratio of this number to the total number of segments in the cluster is calculated to obtain the cluster non-inverse ratio. The cluster inverse ratio and decay inverse value of the spatial cluster are used as inverse parameters. The absolute change in load impact and the negative fluctuation ratio of each weakened section in the spatial cluster are used as impact parameters. A reversible prediction model is constructed based on reversible and impact parameters, and the decay transformation time of all reversibly decayed impact-weakening segments within the spatial cluster into irreversible impact-weakening segments is output.

10. A method for dynamic monitoring and fault pre-diagnosis of multiple parameters of cable insulation, characterized in that: This is used to implement the cable insulation multi-parameter dynamic monitoring and fault pre-diagnosis system as described in any one of claims 1-9.

Images