A power distribution cable micro-power consumption detection method and system

Abstract

The present application relates to power distribution cable state monitoring technical field, specifically disclose a kind of power distribution cable micro-power consumption detection method and system, through the deployment micro-power consumption multi-physical field sensor array synchronous acquisition cable joint current, temperature distribution, mechanical vibration and environmental temperature and humidity signal;Multi-source signal is extracted to obtain load fluctuation characteristic spectrum, temperature gradient distribution characteristics and vibration energy spectrum characteristics by multiscale feature;By fuzzy inference and feature fusion, the feature is mapped to fuzzy state space, and the comprehensive health evaluation index is output based on three-level fuzzy rule inference;Dynamic evolution analysis is carried out to the index, and time series dependent relationship is constructed to predict life attenuation curve and determine optimal maintenance time window;According to the length of time window, generate hierarchical warning information, solve the problem that existing monitoring technology response lag, high and energy consumption, can identify the implicit defect caused by multi-physical field coupling, realize the quantitative evaluation and residual life prediction of cable joint health state.

Description

Technical Field

[0001] This invention relates to the field of power distribution cable condition monitoring technology, specifically to a method and system for detecting low power consumption in power distribution cables. Background Technology

[0002] In smart grid distribution systems, power cables and their connectors, as critical connection components, operate long-term in a complex environment involving electro-thermal-mechanical coupling. Existing technologies primarily rely on periodic inspections, preventative testing, and online monitoring devices based on single-parameter thresholds (such as temperature exceeding limits) for condition management. These methods generally suffer from problems such as response lag and high missed detection rates: manual inspections and periodic testing cannot achieve continuous condition awareness, while conventional online monitoring can only capture overt faults, lacking effective means to identify latent defects such as progressive insulation degradation and slow changes in contact resistance caused by the coupling effects of multiple factors. Furthermore, online monitoring equipment deployed to achieve condition awareness often faces bottlenecks such as high energy consumption and short battery life, making large-scale long-term deployment on passive or remote lines difficult, resulting in a dual dilemma of insufficient coverage and depth in distribution cable condition monitoring.

[0003] The existing technology has the following shortcomings: The system cannot identify "latent overheating" faults in cable joints caused by the coupling of DC bias magnetism induced by space weather disturbances (such as geomagnetic storms) with normal load thermal cycling. The DC component induced by the geomagnetic field causes additional ferromagnetic losses in the cable's metallic sheath. This latent heat source, combined with the thermal effect of the load current, can keep the joint material within its recrystallization critical temperature range for extended periods, leading to metal annealing and creep damage, while the overall joint temperature does not exceed the static alarm threshold of traditional monitoring. This gradual degradation of the material's microstructure caused by multi-physics coupling develops silently even when all single-parameter monitoring data appear "normal," creating a difficult-to-detect and potentially serious technical blind spot in the existing single-parameter threshold monitoring system. Furthermore, existing high-power monitoring solutions cannot provide sustainable data support for identifying such complex faults. Summary of the Invention

[0004] The purpose of this invention is to provide a method and system for detecting low power consumption in power distribution cables, so as to solve the problems mentioned above.

[0005] The objective of this invention can be achieved through the following technical solutions: A method for detecting low power consumption in power distribution cables includes the following steps: S1: Through a low-power multi-physics field sensor array deployed at the cable joint, multi-source signals of the cable body are collected synchronously, including: current signal, temperature distribution signal, mechanical vibration signal and ambient temperature and humidity signal; S2: Perform multi-scale feature extraction on the acquired multi-source signals to obtain a multi-dimensional feature set including load fluctuation feature spectrum, temperature gradient distribution features and vibration energy spectrum features; S3: Perform fuzzy reasoning and feature fusion on the multi-dimensional feature set, map each feature to the fuzzy state space through membership degree calculation, and perform reasoning based on multi-level fuzzy rules to output a comprehensive health assessment index. S4: Perform dynamic evolution analysis on comprehensive health assessment indicators, predict the life decay curve of cable joints by constructing the temporal dependency relationship between historical data and future states, and determine the optimal maintenance time window. S5: Generate and output the corresponding early warning level information based on the optimal maintenance time window.

[0006] As a further aspect of the present invention: the process of obtaining the load fluctuation characteristic spectrum is as follows: Divide the current signal into a sliding window of stacked data according to a preset duration; The initial spectrum is obtained by performing a spectral transformation on the current sequence within each window. Calculate the coefficient of variation of the initial spectrum of each sliding window to form a spectrum variation sequence; Perform convolution operation between the spectral variation sequence and the typical load spectral template; The envelope of the convolution output is extracted as the load fluctuation feature spectrum.

[0007] As a further aspect of the present invention: the process for obtaining the temperature gradient distribution characteristics is as follows: Obtain temperature readings from multiple temperature measurement points on the surface of the cable joint; Construct a temperature distribution matrix based on the spatial location of the temperature values ​​at each temperature measurement point; Calculate the rate of temperature change along both the axial and radial directions of the cable. Perform tensor product operation between the axial temperature change rate and the radial temperature change rate; The temperature gradient distribution characteristics are obtained by performing eigenvalue decomposition on the calculation results.

[0008] As a further aspect of the present invention: the process of obtaining the vibration energy spectrum characteristics is as follows: Overlapping segmentation processing is performed on the vibration signal; Power spectral density estimation is performed on each segment of the signal; Calculate the energy integral value of each segment of the power spectrum in a specific frequency band; The energy integral values ​​of different segments are arranged in chronological order to form an energy sequence; The vibrational energy spectrum characteristics are obtained by performing a difference operation on the energy sequence.

[0009] As a further aspect of the present invention: the step of performing fuzzy reasoning and feature fusion on the multi-dimensional feature set, and mapping each feature to a fuzzy state space through membership degree calculation, specifically includes: Construct nonlinear membership functions targeting load fluctuation characteristic spectrum, temperature gradient distribution characteristics, and vibration energy spectrum characteristics; Input the numerical values ​​of load fluctuation characteristic spectrum, temperature gradient distribution characteristic and vibration energy spectrum characteristic into the corresponding membership function to calculate the membership degree of the three states: normal, attention and abnormal. The inflection point thresholds of each membership function are dynamically adjusted based on the cable's operating status. All calculated membership values ​​are constructed into a fuzzy state vector according to feature categories; Multiple fuzzy state vectors are merged into a unified comprehensive fuzzy state vector through weighted aggregation.

[0010] As a further aspect of the present invention: the reasoning based on multi-level fuzzy rules to output a comprehensive health assessment index specifically includes: Establish a three-level fuzzy rule base comprising a feature layer, a coupling layer, and a decision layer; Single-feature fuzzy rules are applied at the feature layer to perform preliminary reasoning on the fuzzy state of each feature; Multi-feature coupling rules are applied in the coupling layer to analyze the correlation between different feature fuzzy states; The decision-making level synthesizes the reasoning results from the first two levels and applies the health status determination rules. By using the centroid method for defuzzification, the final fuzzy inference result is converted into a precise health assessment indicator.

[0011] As a further aspect of the present invention: the predicted lifespan degradation curve of the cable joint specifically includes: Obtain historical sequences of comprehensive health assessment indicators for multiple consecutive monitoring periods; Construct a sliding time window to extract the dynamic change characteristics of health assessment indicators; A nonlinear decay function is constructed based on the rate of change of health assessment indicators; The characteristic parameters of the decay function are optimized through an iterative correction algorithm. Generate a life decay curve that reflects the changing trend of the remaining life of the cable joint.

[0012] As a further aspect of the present invention: determining the optimal maintenance time window specifically includes: Set reliability thresholds for health assessment indicators; Calculate the time of intersection between the lifetime decay curve and the reliability threshold; Confidence correction is applied to the intersection time based on the risk assessment matrix; The time window boundaries are determined by comprehensively considering maintenance resource factors; The output includes the optimal maintenance time window, which includes the start and end times.

[0013] As a further aspect of the present invention: the step of generating and outputting corresponding early warning level information based on the optimal maintenance time window specifically includes: When the maintenance time window exceeds the first preset threshold, a first-level early warning is generated to maintain the basic monitoring status. When the maintenance time window is between the second preset threshold and the first preset threshold, a second-level early warning is generated, the monitoring frequency is increased, and a preliminary analysis report is generated. When the maintenance time window is between the third preset threshold and the second preset threshold, a third-level warning is generated, multi-parameter joint diagnosis is initiated, and maintenance resources are prepared. When the maintenance time window is less than the third preset threshold, a fourth-level warning is generated, a maintenance notice is immediately pushed out, and the emergency response process is initiated.

[0014] A low-power detection system for power distribution cables, comprising: The multi-physics synchronous acquisition module, through a low-power multi-physics sensor array deployed at the cable joint, synchronously acquires multi-source signals of the cable body, including: current signal, temperature distribution signal, mechanical vibration signal and ambient temperature and humidity signal. The multi-scale feature extraction module is used to extract multi-scale features from the acquired multi-source signals to obtain a multi-dimensional feature set including load fluctuation feature spectrum, temperature gradient distribution features and vibration energy spectrum features. The intelligent feature fusion module is used to perform fuzzy reasoning and feature fusion on a multi-dimensional feature set. It maps each feature to a fuzzy state space through membership degree calculation, and performs reasoning based on multi-level fuzzy rules to output a comprehensive health assessment index. The state evolution prediction module is used to perform dynamic evolution analysis on comprehensive health assessment indicators. By constructing the temporal dependency relationship between historical data and future states, it predicts the life decay curve of cable joints and determines the optimal maintenance time window. The adaptive early warning module generates and outputs corresponding early warning level information based on the optimal maintenance time window.

[0015] The beneficial effects of this invention are: (1) This invention, through multi-physics sensing and multi-scale feature extraction, can capture potential risks that are difficult to detect, such as "hidden overheating" caused by the coupling effect of DC bias and thermal cycling. Furthermore, by utilizing fuzzy reasoning and dynamic evolution analysis, a complete technical chain is constructed from "multi-source data" to "health indicators" and then to "life prediction," realizing the quantitative assessment of the health status of cable joints and the trend prediction of their remaining life. This enables maintenance personnel to plan and execute precise maintenance measures in advance based on the predicted optimal maintenance time window and graded early warning information before a fault occurs, thereby transforming post-fault maintenance into pre-fault prevention, effectively avoiding sudden power outages and improving the power supply reliability of the distribution network.

[0016] (2) Existing online monitoring equipment often operates at full power continuously in pursuit of reliability, resulting in high energy consumption, high maintenance costs, and difficulty in large-scale deployment. This approach dynamically links monitoring strategies to equipment health status, adaptively adjusting warning levels and corresponding monitoring intensity based on the predicted optimal maintenance time window. When equipment is healthy, the system maintains a low-frequency "basic monitoring mode" to achieve extremely low power consumption; as the risk level increases, the system gradually activates "enhanced monitoring" and even "precise diagnosis mode" to ensure high-precision data acquisition at critical moments. This "on-demand allocation" intelligent monitoring strategy optimizes the overall system power consumption to the maximum extent while absolutely guaranteeing the reliability of fault warnings, providing a feasible technical path for deploying online monitoring devices in power distribution networks. Attached Figure Description

[0017] The invention will now be further described with reference to the accompanying drawings.

[0018] Figure 1 This is a flowchart of the method of the present invention; Figure 2 This is a system block diagram of the present invention. Detailed Implementation

[0019] 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.

[0020] Please see Figure 1 As shown, the present invention is a method for detecting low power consumption in power distribution cables, comprising the following steps: S1: By deploying a low-power multi-physics field sensor array at the cable joint, multi-source signals of the cable body are collected synchronously, including: current signal, temperature distribution signal, mechanical vibration signal and ambient temperature and humidity signal. S2: Perform multi-scale feature extraction on the acquired multi-source signals to obtain a multi-dimensional feature set including load fluctuation feature spectrum, temperature gradient distribution features and vibration energy spectrum features; S3: Perform fuzzy reasoning and feature fusion on the multi-dimensional feature set, map each feature to the fuzzy state space through membership degree calculation, and perform reasoning based on multi-level fuzzy rules to output a comprehensive health assessment index. S4: Perform dynamic evolution analysis on comprehensive health assessment indicators, predict the life decay curve of cable joints by constructing the temporal dependency relationship between historical data and future states, and determine the optimal maintenance time window. S5: Generate and output the corresponding early warning level information based on the optimal maintenance time window.

[0021] In S1, a low-power multiphysics sensor array deployed at the cable joint synchronously acquires multi-source signals from the cable body, including: current signals, temperature distribution signals, mechanical vibration signals, and ambient temperature and humidity signals. Specifically, these include: The data acquisition of this invention is achieved through a low-power multiphysics sensor array deployed at the cable joint. The sensor array consists of multiple sensor units dedicated to measuring different physical quantities. Each unit is arranged in space according to a preset topology on the surface of the cable joint and in the adjacent area to ensure the spatial representativeness and synchronization of the acquired data.

[0022] The current signal is acquired using an open-type current sensing unit based on the Hall effect. This unit is directly mounted on the periphery of the conductor connected to the cable connector. By detecting changes in the intensity of the magnetic field around the conductor, a voltage signal proportional to the load current is obtained. This non-contact measurement method does not affect the normal conductivity of the cable and simultaneously achieves electrical isolation.

[0023] Temperature distribution signals are acquired using a miniature infrared thermometer array deployed on the surface of the cable joint. This array consists of multiple temperature-sensitive cells arranged in a grid pattern, each cell independently measuring the surface temperature at its corresponding location. By reading the measurements from all the temperature-sensitive cells, two-dimensional temperature field distribution data of the cable joint surface is constructed.

[0024] Mechanical vibration signals are acquired using a microelectromechanical system (MEMS) accelerometer unit. This unit is fixed to the mechanical vibration transmission path of the cable connector housing using a dedicated clamp, directly sensing the vibration acceleration of the cable connector in three-dimensional space. The output signal of the sensor unit reflects the mechanical vibration state of the connector caused by internal loosening, external excitation, or electrodynamic forces.

[0025] The acquisition of ambient temperature and humidity signals is achieved through a temperature and humidity composite sensing unit installed on the support structure near the cable joint. This unit is located outside the joint at an appropriate distance and is used to measure the air temperature and relative humidity parameters of the local environment where the joint is located, providing an environmental condition reference for subsequent analysis.

[0026] All the above sensing units are synchronously sampled using a unified clock signal to ensure that the collected data of all physical quantities have the same time reference. The collected multi-source signals are transmitted to the local processing unit via fieldbus to form a complete dataset containing current, temperature distribution, mechanical vibration, and ambient temperature and humidity.

[0027] In S2, multi-scale feature extraction is performed on the acquired multi-source signals to obtain a multi-dimensional feature set including load fluctuation feature spectrum, temperature gradient distribution features, and vibration energy spectrum features, specifically including: The multi-scale feature extraction process of this invention includes three main parts: load fluctuation feature spectrum extraction, temperature gradient distribution feature extraction, and vibration energy spectrum feature extraction.

[0028] The extraction process of the load fluctuation characteristic spectrum is as follows: First, the continuously acquired current signal is divided into multiple overlapping sliding windows of fixed duration, each containing the same number of sampling points. Spectral analysis is performed on the current sequence within each window, transforming it from the time domain to the frequency domain to obtain an initial spectrum containing the energy of each frequency component. Next, the coefficient of variation of the initial spectrum within each sliding window is calculated. This coefficient is obtained by dividing the standard deviation of the spectral values ​​by the mean, forming a spectral variation sequence reflecting the spectral fluctuation characteristics. Then, this sequence is convolved with a pre-established typical load spectrum template, which is a typical spectral pattern obtained through statistical analysis of historical normal load data. Finally, the envelope of the convolution output sequence is extracted. This envelope is obtained by connecting all local extrema and smoothing it, serving as the load fluctuation characteristic spectrum.

[0029] The extraction process of temperature gradient distribution features is as follows: First, temperature readings are obtained from multiple temperature measuring points arranged in a regular grid pattern on the surface of the cable joint. The temperature values ​​of each measuring point are then used to construct a two-dimensional temperature distribution matrix based on their spatial relationship, with each element corresponding to the temperature value of a measuring point. The temperature change rate is calculated along both the axial and radial directions of the cable. The axial temperature change rate is obtained by calculating the ratio of the temperature difference between adjacent measuring points to their distance, while the radial temperature change rate is obtained by calculating the temperature gradient between the inner and outer measuring points on the same cross-section. The axial and radial temperature change rate matrices are then multiplied by a tensor, which combines all elements of the two matrices pairwise into a new higher-order tensor. Finally, the tensor multiplication result is decomposed into eigenvalues, and the largest eigenvalues ​​are extracted as the temperature gradient distribution features.

[0030] The extraction process of vibration energy spectrum features is as follows: First, the continuously acquired vibration signals are segmented by overlap, dividing the signal into multiple overlapping time periods. Power spectral density estimation is performed on the vibration signal of each segment, and the power distribution of each frequency component is calculated using an improved periodogram method. The energy integral value of the power spectrum of each segment in a specific frequency band is calculated. The specific frequency band is a frequency range predetermined based on the typical fault vibration characteristics of cable joints. The energy integral is obtained by summing the power values ​​at all frequency points within that frequency band. The energy integral values ​​calculated for different time periods are arranged in chronological order to form a vibration energy sequence. Finally, a difference operation is performed on this energy sequence to calculate the difference in energy values ​​between adjacent time points, obtaining the vibration energy spectrum features reflecting the trend of vibration energy change.

[0031] In S3, fuzzy reasoning and feature fusion are performed on the multi-dimensional feature set. Membership degree calculation maps each feature to a fuzzy state space, and reasoning is performed based on multi-level fuzzy rules to output a comprehensive health assessment index, specifically including: The process of mapping a multidimensional feature set to a fuzzy state space is as follows: First, nonlinear membership functions are constructed for the load fluctuation feature spectrum, temperature gradient distribution feature, and vibration energy spectrum feature. These membership functions adopt an S-shaped curve form, and their shape is defined by three key parameters: the starting point, the inflection point, and the ending point. The starting point represents the minimum value at which the feature value begins to belong to a certain state, the inflection point represents the value at which the feature value is most likely to belong to that state, and the ending point represents the maximum value at which the feature value no longer belongs to that state. For each feature, membership functions corresponding to the three states of normal, attention, and abnormal are constructed respectively.

[0032] The numerical values ​​of each feature are input into the corresponding membership function to calculate its membership degree to the three states. The calculation process for membership degree is as follows: based on the vertical coordinate value corresponding to the input feature value on the membership function curve, this value is between 0 and 1, indicating the degree of belonging to that state. For example, when the value of the load fluctuation characteristic spectrum is located at the inflection point of the normal state, its membership degree to the normal state is 1, and its membership degree to the attention state and the abnormal state is 0.

[0033] The inflection point thresholds of each membership function are dynamically adjusted based on the cable's operating status. The adjustment process is based on historical cable operating data and current environmental conditions, using statistical analysis to determine the normal fluctuation range of each characteristic value. When the ambient temperature is high, the normal state thresholds of temperature-related characteristics are appropriately increased; when the load is heavy, the evaluation criteria for load characteristics are adjusted accordingly.

[0034] All calculated membership values ​​are constructed into fuzzy state vectors according to feature categories. Each feature corresponds to a three-dimensional vector, with the three dimensions representing the degree to which the feature belongs to the normal, attentive, and abnormal states, respectively. The fuzzy state vectors of all features form a multi-dimensional vector set, which fully describes the fuzzy state distribution of the cable joint at the current moment.

[0035] Multiple fuzzy state vectors are merged into a unified comprehensive fuzzy state vector through weighted aggregation. The weight allocation is determined based on the degree of influence of each feature on the health status of the cable joint: load fluctuation feature has a weight of 0.4, temperature gradient distribution feature has a weight of 0.35, and vibration energy spectrum feature has a weight of 0.25. Weighted aggregation is achieved by weighted summation of the corresponding dimensions of each vector, ultimately resulting in a three-dimensional comprehensive fuzzy state vector.

[0036] The reasoning process based on multi-level fuzzy rules is as follows: First, a three-level fuzzy rule base is established, comprising a feature layer, a coupling layer, and a decision layer. The feature layer rules take the form "If feature A belongs to state X, then the preliminary conclusion is Y," containing 9 rules covering all combinations of the three features in the three states. The coupling layer rules consider the interaction between features, taking the form "If feature A belongs to state X and feature B belongs to state Y, then the coupling effect is Z," containing 27 rules. The decision layer rules synthesize the conclusions of the first two layers, taking the form "If the feature layer conclusion is A and the coupling layer conclusion is B, then the final state is C," containing 10 rules.

[0037] Single-feature fuzzy rules are applied at the feature layer to perform preliminary reasoning on the fuzzy state of each feature. The reasoning process employs the min-max reasoning method, taking the smaller value between the membership degree of the precondition and the rule weight, and then taking the larger value for the output of all rules. After each feature undergoes inference at the feature layer, a preliminary state evaluation result for that feature is obtained.

[0038] Multi-feature coupling rules are applied at the coupling layer to analyze the correlation between different feature fuzzy states. These rules specifically consider important coupling phenomena such as the mutual reinforcement effect of temperature and vibration, and the coordinated change relationship between load and temperature. Intermediate evaluation results considering the interactions between features are obtained through coupling layer inference.

[0039] The decision-making layer synthesizes the reasoning results from the first two layers and applies health status determination rules. These rules comprehensively consider the outputs of the feature layer and the coupling layer, combined with the operating mechanism of the cable joint, to derive the final fuzzy set of health statuses. The weights of the decision-making layer rules are determined based on actual operational experience and fault statistics.

[0040] The centroid method is used for defuzzification, transforming the final fuzzy inference result into a precise health assessment indicator. The centroid method calculation process involves taking the centroid of the area enclosed by the membership function curve of the output fuzzy set and the horizontal axis as the precise health assessment indicator. This indicator ranges from 0 to 100, with higher values ​​indicating better health.

[0041] In S4, dynamic evolution analysis is performed on the comprehensive health assessment indicators. By constructing the temporal dependency relationship between historical data and future states, the life decay curve of the cable joint is predicted, and the optimal maintenance time window is determined. Specifically, this includes: The prediction process for the lifespan decline curve is as follows: First, obtain historical sequences of comprehensive health assessment indicators for multiple consecutive monitoring periods. The monitoring period is set to 24 hours, and data for no less than 90 consecutive periods are acquired. The historical sequences are arranged in chronological order, and each data point includes a timestamp and the corresponding health assessment indicator value. Data preprocessing includes removing obvious outliers and supplementing missing data. Missing data is supplemented using linear interpolation of adjacent data.

[0042] A sliding time window was constructed to extract the dynamic characteristics of health assessment indicators. The sliding window size was set to 30 periods, with each sliding step being one period. Within each window, the mean, variance, coefficient of variation, and trend of the health assessment indicators were calculated. The trend was obtained through linear regression analysis, with the regression coefficient representing the average rate of change of the health assessment indicators. The sliding coefficient of variation was calculated as the ratio of the standard deviation to the mean, reflecting the relative volatility of the health assessment indicators.

[0043] A nonlinear decay function is constructed based on the rate of change of health assessment indicators. The decay function adopts an exponential decay form and includes three parameters: initial health state, decay coefficient, and time variable. The initial health state is taken as the current health assessment indicator value, and the decay coefficient is determined based on the changing trend of the most recent 30 cycles. Specifically, the nonlinear decay function is expressed as: health state equals the product of the initial health state multiplied by the base of the natural logarithm, the negative decay coefficient, and the time variable raised to the power of the time variable. The time variable is in days, representing the prediction duration from the current moment.

[0044] The characteristic parameters of the attenuation function are optimized through an iterative correction algorithm. The iterative process employs gradient descent, using the mean squared error between the predicted and actual values ​​as the loss function. In each iteration, the partial derivatives of the loss function with respect to each parameter are calculated, and the parameters are updated in the reverse gradient direction at a learning rate of 0.01. The iteration terminates when the change in the loss function is less than 0.001 after 10 consecutive iterations. The optimized parameters more accurately reflect the actual attenuation pattern of the cable joint.

[0045] Generate a lifespan decay curve reflecting the changing trend of the remaining lifespan of the cable joint. Extend the optimized decay function along the time dimension to plot a complete curve from the current moment until the health assessment index drops to a predetermined threshold. The horizontal axis of the curve represents time, and the vertical axis represents the predicted health assessment index value. The curve provides a visual understanding of the future trend of the cable joint's health status, offering a basis for maintenance decisions.

[0046] The process for determining the optimal maintenance time window is as follows: First, a reliability threshold for the health assessment indicators is set. This reliability threshold is determined based on the cable joint's design life, operating environment, and importance level, and is typically set to 60. This threshold represents the minimum health condition required for the cable joint to maintain reliable operation; values ​​below this threshold indicate a higher risk of failure.

[0047] Calculate the intersection time of the lifetime degradation curve and the reliability threshold. On the lifetime degradation curve, find the earliest time point where the health assessment index value equals the reliability threshold. This time point represents the moment when the cable joint's health condition is expected to drop below the reliability threshold. The calculation process uses a binary search method, gradually narrowing the search interval within the curve's time range until a time point meeting the accuracy requirements is found.

[0048] The confidence level of the intersection time is adjusted based on a risk assessment matrix. The risk assessment matrix includes two dimensions: the volatility and the trend of the health assessment indicators. Volatility is categorized into low, medium, and high levels, while the trend is categorized into slowing, stable, and accelerating levels. The actual level is determined based on data analysis from the most recent 30 periods, and the corresponding confidence adjustment coefficient is obtained from a table. The confidence adjustment coefficient ranges from 0.8 to 1.2 and is used to adjust the conservatism of the intersection time.

[0049] The time window boundaries are determined by comprehensively considering maintenance resource factors, including spare parts inventory, personnel arrangements, and power outage plans. The start time of the time window is the intersection of the two time points minus the preparation time, which is determined to be 7 to 14 days depending on the complexity of the maintenance. The end time of the time window is the intersection of the two time points plus a buffer time, which is determined to be 3 to 7 days depending on the difficulty of allocating maintenance resources. This determined time window takes into account both technical conditions and practical operational feasibility.

[0050] The output includes the optimal maintenance time window, containing the start and end times. The time window is represented in date format, accurate to the day. The output information includes the specific time interval and the corresponding confidence level, which is calculated using a confidence correction factor, ranging from 0 to 100%.

[0051] In S5, based on the optimal maintenance time window, corresponding early warning level information is generated and output, specifically including: When the maintenance time window exceeds the first preset threshold, a first-level warning is generated to maintain the basic monitoring status. The basic monitoring status maintains the original monitoring parameters and sampling frequency, with a data collection cycle of 24 hours. The monitoring data is only used for daily recording and trend analysis and does not trigger additional processing procedures.

[0052] When the maintenance time window falls between the second preset threshold and the first preset threshold, a second-level warning is generated, the monitoring frequency is increased, and a preliminary analysis report is generated. The monitoring frequency is adjusted to collect complete data once every 6 hours, and the analysis report includes the changing trend of health assessment indicators, analysis of major influencing factors, and preliminary maintenance procedures.

[0053] When the maintenance time window is between the third preset threshold and the second preset threshold, a third-level warning is generated, multi-parameter joint diagnosis is initiated, and maintenance resources are prepared. The multi-parameter joint diagnosis simultaneously analyzes the correlation characteristics of electrical, thermal, and mechanical parameters. Maintenance resource preparation includes the arrangement of maintenance personnel, allocation of spare parts, and the formulation of a preliminary power outage plan.

[0054] When the maintenance time window is less than the third preset threshold, a fourth-level warning is generated, a maintenance notice is immediately pushed out and the emergency response process is initiated. The maintenance notice is sent to relevant maintenance personnel and management personnel through preset information channels. The emergency response process includes on-site inspection arrangements, implementation of temporary protection measures and activation of the fault emergency plan.

[0055] Please see Figure 2 As shown, a low-power detection system for power distribution cables includes: The multi-physics synchronous acquisition module, through a low-power multi-physics sensor array deployed at the cable joint, synchronously acquires multi-source signals of the cable body, including: current signal, temperature distribution signal, mechanical vibration signal and ambient temperature and humidity signal. The multi-scale feature extraction module is used to extract multi-scale features from the acquired multi-source signals to obtain a multi-dimensional feature set including load fluctuation feature spectrum, temperature gradient distribution features and vibration energy spectrum features. The intelligent feature fusion module is used to perform fuzzy reasoning and feature fusion on a multi-dimensional feature set. It maps each feature to a fuzzy state space through membership degree calculation, and performs reasoning based on multi-level fuzzy rules to output a comprehensive health assessment index. The state evolution prediction module is used to perform dynamic evolution analysis on comprehensive health assessment indicators. By constructing the temporal dependency relationship between historical data and future states, it predicts the life decay curve of cable joints and determines the optimal maintenance time window. The adaptive early warning module generates and outputs corresponding early warning level information based on the optimal maintenance time window.

[0056] The working principle of this invention is as follows: A low-power multi-physics field sensor array deployed at the cable joint synchronously collects multi-source signals, including current signals, temperature distribution signals, mechanical vibration signals, and ambient temperature and humidity signals. Then, multi-scale feature extraction is performed on the collected multi-source signals to obtain a multi-dimensional feature set containing load fluctuation characteristic spectrum, temperature gradient distribution characteristics, and vibration energy spectrum characteristics. Next, fuzzy reasoning and feature fusion are performed on the multi-dimensional feature set. Membership degree calculation maps each feature to a fuzzy state space, and reasoning is performed based on a three-level fuzzy rule system containing a feature layer, a coupling layer, and a decision layer to output a comprehensive health assessment index. Then, dynamic evolution analysis is performed on the comprehensive health assessment index. By constructing the temporal dependency relationship between historical data and future states, the lifespan decay curve of the cable joint is predicted, and the optimal maintenance time window is determined. Finally, based on the length of the optimal maintenance time window, corresponding early warning level information is generated and output, realizing graded early warning from normal monitoring to emergency response.

[0057] 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 claims of this invention should still fall within the patent coverage of this invention.

Claims

1. A method for detecting low power consumption in power distribution cables, characterized in that, Includes the following steps: S1: By deploying a low-power multi-physics field sensor array at the cable joint, multi-source signals of the cable body are collected synchronously, including: current signal, temperature distribution signal, mechanical vibration signal and ambient temperature and humidity signal. S2: Perform multi-scale feature extraction on the acquired multi-source signals to obtain a multi-dimensional feature set including load fluctuation feature spectrum, temperature gradient distribution features and vibration energy spectrum features; S3: Perform fuzzy reasoning and feature fusion on the multi-dimensional feature set, map each feature to the fuzzy state space through membership degree calculation, and perform reasoning based on multi-level fuzzy rules to output a comprehensive health assessment index. S4: Perform dynamic evolution analysis on comprehensive health assessment indicators, predict the life decay curve of cable joints by constructing the temporal dependency relationship between historical data and future states, and determine the optimal maintenance time window. S5: Generate and output the corresponding early warning level information based on the optimal maintenance time window.

2. The method for detecting low power consumption in power distribution cables according to claim 1, characterized in that, The process of obtaining the load fluctuation characteristic spectrum is as follows: Divide the current signal into a sliding window of stacked data according to a preset duration; The initial spectrum is obtained by performing a spectral transformation on the current sequence within each sliding window. Calculate the coefficient of variation of the initial spectrum of each sliding window to form a spectrum variation sequence; Perform convolution operation between the spectral variation sequence and the typical load spectral template; The envelope of the convolution output is extracted as the load fluctuation feature spectrum.

3. The method for detecting low power consumption in power distribution cables according to claim 1, characterized in that, The process of obtaining the temperature gradient distribution characteristics is as follows: Obtain temperature readings from multiple temperature measurement points on the surface of the cable joint; Construct a temperature distribution matrix based on the spatial location of the temperature values ​​at each temperature measurement point; Calculate the rate of temperature change along both the axial and radial directions of the cable. Perform tensor product operation between the axial temperature change rate and the radial temperature change rate; The temperature gradient distribution characteristics are obtained by performing eigenvalue decomposition on the calculation results.

4. The method for detecting low power consumption in power distribution cables according to claim 1, characterized in that, The process of obtaining the vibrational energy spectrum characteristics is as follows: Overlapping segmentation processing is performed on mechanical vibration signals; Power spectral density estimation is performed on each segment of the signal; Calculate the energy integral value of each segment of the power spectrum in a specific frequency band; The energy integral values ​​of different segments are arranged in chronological order to form an energy sequence; The vibrational energy spectrum characteristics are obtained by performing a difference operation on the energy sequence.

5. The method for detecting low power consumption in power distribution cables according to claim 1, characterized in that, The process of performing fuzzy reasoning and feature fusion on a multi-dimensional feature set, and mapping each feature to a fuzzy state space through membership degree calculation, specifically includes: Construct nonlinear membership functions targeting load fluctuation characteristic spectrum, temperature gradient distribution characteristics, and vibration energy spectrum characteristics; Input the numerical values ​​of load fluctuation characteristic spectrum, temperature gradient distribution characteristic and vibration energy spectrum characteristic into the corresponding membership function to calculate the membership degree of the three states: normal, attention and abnormal. The inflection point thresholds of each membership function are dynamically adjusted based on the cable's operating status. All calculated membership values ​​are constructed into a fuzzy state vector according to feature categories; Multiple fuzzy state vectors are merged into a unified comprehensive fuzzy state vector through weighted aggregation.

6. The method for detecting low power consumption in power distribution cables according to claim 1, characterized in that, The reasoning based on multi-level fuzzy rules outputs a comprehensive health assessment index, specifically including: Establish a three-level fuzzy rule base comprising a feature layer, a coupling layer, and a decision layer; Single-feature fuzzy rules are applied at the feature layer to perform preliminary reasoning on the fuzzy state of each feature; Multi-feature coupling rules are applied in the coupling layer to analyze the correlation between different feature fuzzy states; The decision-making level synthesizes the reasoning results from the first two levels and applies the health status determination rules. By using the centroid method for defuzzification, the final fuzzy inference result is converted into a precise health assessment indicator.

7. The method for detecting low power consumption in power distribution cables according to claim 1, characterized in that, The predicted lifespan degradation curve of the cable joint specifically includes: Obtain historical sequences of comprehensive health assessment indicators for multiple consecutive monitoring periods; Construct a sliding time window to extract the dynamic change characteristics of comprehensive health assessment indicators; A nonlinear decay function is constructed based on the dynamic rate of change of comprehensive health assessment indicators; The characteristic parameters of the decay function are optimized through an iterative correction algorithm. Generate a life decay curve that reflects the changing trend of the remaining life of the cable joint.

8. The method for detecting low power consumption in power distribution cables according to claim 1, characterized in that, The determination of the optimal maintenance time window specifically includes: Set reliability thresholds for comprehensive health assessment indicators; Calculate the time of intersection between the lifetime decay curve and the reliability threshold; Confidence correction is applied to the intersection time based on the risk assessment matrix; The time window boundaries are determined by comprehensively considering maintenance resource factors; The output includes the optimal maintenance time window, which includes the start and end times.

9. The method for detecting low power consumption in power distribution cables according to claim 1, characterized in that, The process of generating and outputting corresponding early warning level information based on the optimal maintenance time window specifically includes: When the maintenance time window exceeds the first preset threshold, a first-level early warning is generated to maintain the basic monitoring status. When the maintenance time window is between the second preset threshold and the first preset threshold, a second-level early warning is generated, the monitoring frequency is increased, and a preliminary analysis report is generated. When the maintenance time window is between the third preset threshold and the second preset threshold, a third-level warning is generated, multi-parameter joint diagnosis is initiated, and maintenance resources are prepared. When the maintenance time window is less than the third preset threshold, a fourth-level warning is generated, a maintenance notice is immediately pushed out, and the emergency response process is initiated.

10. A low-power detection system for power distribution cables, characterized in that, A method for detecting low power consumption of a power distribution cable according to any one of claims 1-9 includes: The multi-physics synchronous acquisition module, through a low-power multi-physics sensor array deployed at the cable joint, synchronously acquires multi-source signals of the cable body, including: current signal, temperature distribution signal, mechanical vibration signal and ambient temperature and humidity signal. The multi-scale feature extraction module is used to extract multi-scale features from the acquired multi-source signals to obtain a multi-dimensional feature set including load fluctuation feature spectrum, temperature gradient distribution features and vibration energy spectrum features. The intelligent feature fusion module is used to perform fuzzy reasoning and feature fusion on a multi-dimensional feature set. It maps each feature to a fuzzy state space through membership degree calculation, and performs reasoning based on multi-level fuzzy rules to output a comprehensive health assessment index. The state evolution prediction module is used to perform dynamic evolution analysis on comprehensive health assessment indicators. By constructing the temporal dependency relationship between historical data and future states, it predicts the life decay curve of cable joints and determines the optimal maintenance time window. The adaptive early warning module generates and outputs corresponding early warning level information based on the optimal maintenance time window.

Images