Magnetic field traveling wave signal processing method for cable fault detection and related equipment

Abstract

The invention provides a magnetic field traveling wave signal processing method for cable fault detection and related equipment, and relates to the technical field of cable fault detection, and the method comprises the steps: configuring a distributed sensor according to an identified key position of a cable, and collecting a magnetic field traveling wave signal of the cable through the distributed sensor, and obtaining a time sequence signal data set; performing preprocessing and signal enhancement on the time sequence signal data set in sequence, and inputting signal features of the time sequence signal data set after signal enhancement into a pre-optimized fault identification model to obtain a fault positioning result of the cable; wherein the fault identification model is obtained by performing model parameter optimization by adopting a pre-established multi-objective function based on a particle swarm and a preset particle search strategy, and the particle search strategy is set according to a preset speed updating mechanism and a current noise interference state. Therefore, the method effectively enhances the accuracy and practicability of the fault detection system while improving the quality of the cable fault signal.

Description

Technical Field

[0001] The present application relates to the technical field of cable fault detection, and in particular to a method for processing magnetic field traveling wave signals for cable fault detection and related equipment. Background Art

[0002] As cable usage continues to expand and their service life increases, cable faults are becoming increasingly prominent. These problems, such as insulation aging, mechanical damage, and localized overheating, can not only lead to power outages but also serious safety incidents such as fires. In practical applications, cables operate in complex environments, and magnetic field traveling wave signals are susceptible to external electromagnetic interference, signal attenuation, and noise. This results in poor signal quality and difficulty in feature extraction, which in turn affects the accuracy and reliability of fault detection.

[0003] Existing cable fault detection systems generally lack dynamic adaptability in terms of sensor configuration, signal processing, and fault identification. This makes it difficult to fully cover critical cable locations and flexibly adjust acquisition parameters based on the cable's operating status, resulting in incomplete or redundant signal acquisition. Signal processing also struggles to effectively suppress noise interference, reducing the performance of the fault identification model. The fault identification model itself struggles to adapt to the complex and changing cable operating environment, resulting in insufficient fault location accuracy. Summary of the Invention

[0004] The purpose of this application is to solve at least one of the above technical deficiencies, especially the technical defect in the prior art that poor quality of collected signals leads to insufficient accuracy in cable fault detection.

[0005] In a first aspect, the present application provides a method for processing a magnetic field traveling wave signal for cable fault detection, the method comprising:

[0006] Based on the identified key locations of the cable, distributed sensors are configured to collect the cable's magnetic field traveling wave signals through the distributed sensors to obtain a time series signal dataset.

[0007] The time series signal dataset is preprocessed and signal enhanced in sequence, and the signal features of the enhanced time series signal dataset are input into a pre-optimized fault identification model to obtain the cable fault location result;

[0008] Among them, the fault identification model is based on a particle swarm and a preset particle search strategy, and is obtained by optimizing the model parameters using a pre-established multi-objective function. The particle search strategy is set according to a preset speed update mechanism and the current noise interference state.

[0009] In one embodiment, the process of identifying the key locations of the cable includes:

[0010] Extract historical fault information of cables and establish fault hotspot features;

[0011] Obtain the cable load operation task execution results and establish the cable abnormal load characteristics;

[0012] Extract the branching position, cross position and interface position of the cable to construct the cable structural features;

[0013] Determine the critical location of the cable based on the fault hotspot characteristics, cable abnormal load characteristics and cable structure characteristics.

[0014] In one embodiment, the process of preprocessing the time series signal data set includes:

[0015] Filter the time series signal data set using a filter;

[0016] Perform interpolation and filling on the filtered time series signal data set;

[0017] Perform signal normalization on the interpolated and padded time series signal dataset.

[0018] In one embodiment, the optimization process of the fault identification model includes:

[0019] Establishing a particle swarm in an initial state according to the initial model parameters of the fault identification model, and determining the current position and current speed of each particle in the particle swarm;

[0020] In each round of iteration, after determining the current noise interference state corresponding to the current particle swarm, the inertia weight of the current particle swarm is adjusted, and the current position and current velocity of each particle in the current particle swarm are updated. The fitness of the updated particle swarm is evaluated according to the multi-objective function until the fitness evaluation result of the updated particle swarm meets the preset update end condition, and the optimal model parameters of the fault recognition model are obtained.

[0021] In one embodiment, the step of determining the current noise interference state corresponding to the current particle swarm includes:

[0022] Establishing an external environment feature set based on the collected external environment data of the cable;

[0023] Perform multi-dimensional noise quantification on the external environment feature set to obtain the external noise interference score;

[0024] Perform noise anomaly evaluation on the signal characteristics corresponding to each particle in the current particle swarm to obtain the current signal noise interference score;

[0025] The external noise interference score and the current signal noise interference score are used as the current noise interference state.

[0026] In one embodiment, the step of updating the current position and current velocity of each particle in the current particle swarm includes:

[0027] According to the number of iterations of the current particle swarm, the current benchmark iteration speed is established;

[0028] Determine the current tracking update speed based on the previous round position and historical optimal position of each particle in the current particle swarm, as well as the global optimal position of the current particle swarm;

[0029] The current position and current velocity of each particle in the current particle swarm are determined according to the current benchmark iteration velocity, the current tracking update velocity, and the adjusted inertia weight.

[0030] In one embodiment, the process of establishing the multi-objective function includes:

[0031] Analyze cable historical fault data and cable normal operation data to extract cable signal characteristics corresponding to different fault stages;

[0032] According to the cable signal characteristics corresponding to each fault stage, initial weights are assigned to various evaluation targets, including positioning accuracy target, signal accuracy target, calculation speed target, and real-time target.

[0033] A time constraint for adjusting weights over time is constructed, and the weights of various evaluation objectives are adjusted to obtain a multi-objective function.

[0034] In a second aspect, the present application provides a magnetic field traveling wave signal processing device for cable fault detection, the device comprising:

[0035] A time series signal data set determination module is used to configure distributed sensors according to the identified key positions of the cable, and collect the cable's magnetic field traveling wave signal through the distributed sensors to obtain a time series signal data set;

[0036] The fault location result determination module is used to preprocess and enhance the time series signal data set in sequence, and input the signal characteristics of the enhanced time series signal data set into a pre-optimized fault identification model to obtain the cable fault location result;

[0037] Among them, the fault identification model is based on a particle swarm and a preset particle search strategy, and is obtained by optimizing the model parameters using a pre-established multi-objective function. The particle search strategy is set according to a preset speed update mechanism and the current noise interference state.

[0038] In a third aspect, the present application provides a storage medium: the storage medium stores computer-readable instructions, and when the computer-readable instructions are executed by one or more processors, the one or more processors execute the steps of the magnetic field traveling wave signal processing method for cable fault detection as described in any of the above embodiments.

[0039] In a fourth aspect, the present application provides a computer device, comprising: one or more processors, and a memory;

[0040] The memory stores computer-readable instructions, and when the computer-readable instructions are executed by one or more processors, the steps of the method for processing a magnetic field traveling wave signal for cable fault detection in any one of the above embodiments are performed.

[0041] It can be seen from the above technical solutions that the embodiments of the present application have the following advantages:

[0042] The magnetic field traveling wave signal processing method and related equipment for cable fault detection provided by this application, by rationally configuring distributed sensors according to the identified key positions of the cable, specifically enhances the signal coverage of the key areas, effectively avoids invalid acquisition and signal loss, and improves the integrity and representativeness of the time series signal data; further, the preprocessing and signal enhancement strategies are adopted to significantly suppress the influence of external electromagnetic interference and noise on the magnetic field traveling wave signal, improve the clarity and feature distinguishability of the signal, and provide a better signal foundation for feature extraction and fault identification; in the fault identification stage, a model parameter adjustment mechanism based on particle swarm optimization is introduced, combined with a particle search strategy considering the noise interference state and an optimization path design of a multi-objective function, so that the model can still maintain high robustness and recognition accuracy under complex working conditions. As a result, this method effectively enhances the accuracy and practicality of the fault detection system while improving the quality of the cable fault signal. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0044] Figure 1 A flowchart of a method for processing a traveling magnetic wave signal for cable fault detection provided in an embodiment of the present application;

[0045] Figure 2 A schematic structural diagram of a magnetic field traveling wave signal processing device for cable fault detection provided in an embodiment of the present application;

[0046] Figure 3 A schematic diagram of the internal structure of a computer device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0047] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0048] This application provides a method for processing magnetic field traveling wave signals for cable fault detection. The following embodiments are described using the method applied to a computer device as an example. It is understood that the computer device can be any device with data processing capabilities, including but not limited to a single server, a server cluster, a personal laptop computer, a desktop computer, etc. Figure 1 As shown, the method includes:

[0049] S101: Distributed sensors are configured according to the identified key positions of the cable, and the distributed sensors are used to collect the traveling wave signals of the magnetic field of the cable to obtain a time series signal data set.

[0050] Critical locations refer to specific areas within a cable identified through multi-source data analysis that are structurally weak, prone to failure, or significantly impact signal transmission. This multi-source data analysis can be based on historical cable fault records, cable design drawings, cable installation structure information, and operating parameters. Critical locations primarily include cable joints, bends, and interfaces between cables and electrical equipment, such as transformers or switchgear. These locations are more susceptible to failure due to process variations, mechanical stress, or electromagnetic interference. Distributed sensors are magnetic field signal acquisition units deployed and coordinated at multiple locations along a cable to detect magnetic field changes around the cable. Distributed sensors offer high sensitivity, support dynamic parameter adjustment, are resistant to interference, and are capable of multi-point simultaneous sampling, making them suitable for capturing traveling wave signals during cable operation with high precision. A time series signal dataset is a collection of traveling wave magnetic field signals recorded by distributed sensors during acquisition. These signals are organized chronologically to form a multidimensional data sequence with elements such as timestamps, sensor numbers, sample values, and cable status tags. This data sequence fully reflects the electromagnetic behavior of the cable over a specific time period.

[0051] Specifically, computer equipment analyzes historical cable operating data and structural information to identify critical cable locations. This analysis process can identify high-prone areas by reading operational data from a SCADA system, cable fault recorder, or online monitoring platform. Combined with information such as the cable's design drawings, laying pattern, bend structure, and branch joint locations, it can infer areas of structural stress concentration or insulation weakness. Furthermore, non-destructive methods such as infrared thermal imaging and partial discharge detection can be used to perform real-time diagnosis of the cable's current operating status, identifying potential fault points and mapping them as critical locations.

[0052] Next, the computer equipment configures distributed sensors along the cable based on the key location identification results. Specifically, magnetic field sensors with fast response speed and high sensitivity are deployed at the cable joints to capture local magnetic field changes caused by poor contact or insulation degradation. In the cable bending area, the sensor deployment density is adjusted according to the turning angle and curvature radius to ensure sufficient signal acquisition coverage in the bending section. At the location connected to the power equipment, sensor models with strong anti-electromagnetic interference capabilities are used to improve the signal stability of such areas in complex electromagnetic environments. By differentiating the configuration of sensor type, deployment location and layout density, it is ensured that magnetic field changes at key locations can be collected in a timely and accurate manner.

[0053] The computer then dynamically configures the sensor's sampling parameters based on the cable's actual operating status. For example, during normal operation, when the cable current is stable and the temperature is suitable, the sensor can be set to medium-to-low frequency sampling to control data volume and ensure basic status recording. When a short circuit, sudden current change, or abnormal local temperature rise is detected, the sensor immediately switches to high-frequency sampling mode, activating the signal trigger mechanism, increasing the sampling frequency, and enabling noise suppression to fully preserve key signal changes during the fault transient. When carrying out temporary capacity expansion tasks, the sensor enters a long-term continuous sampling state and simultaneously collects parameters such as ambient temperature and humidity to facilitate comprehensive multi-dimensional status analysis. This allows for flexible adjustments based on the cable's real-time operating conditions, improving the sensitivity of detecting faults or abnormal events. The following table shows exemplary distributed sensor parameter configuration information for different cable states.

[0054]

[0055] Finally, a computer integrates the magnetic field traveling wave signals collected by each sensor in chronological order to construct a standardized time-series signal dataset. This dataset records information such as the signal's occurrence time, source location, waveform data, and cable operating status, forming a complete set of multi-channel, traceable, and clearly structured input data that can be used to train fault identification models or directly for real-time judgment and analysis.

[0056] It's understandable that by identifying and focusing coverage on critical locations within the cable, the sensor's coverage of fault-sensitive areas can be significantly improved. Furthermore, dynamically configuring sampling parameters based on the cable's real-time operating status helps reduce invalid data generation under normal conditions and improves sampling accuracy and timeliness during faults or sudden changes. This results in a time series signal dataset with strong representativeness, a high signal-to-noise ratio, and reasonable information density.

[0057] S102: Preprocessing and signal enhancement are performed on the time series signal data set in sequence, and the signal features of the enhanced time series signal data set are input into a pre-optimized fault identification model to obtain a cable fault location result.

[0058] Among them, the fault identification model is based on a particle swarm and a preset particle search strategy, and is obtained by optimizing the model parameters using a pre-established multi-objective function. The particle search strategy is set according to a preset speed update mechanism and the current noise interference state.

[0059] Signal enhancement refers to increasing the expression strength of key signal features through noise suppression, feature highlighting, and amplitude normalization, based on signal preprocessing. This enhances signal discriminability and recognition stability, thereby optimizing subsequent model input. The fault recognition model is a classification model developed by extracting and training features from historical cable signals. It can determine the cable's operating status and identify fault types based on input signal characteristics. During training, the model incorporates a particle swarm algorithm for parameter optimization to improve model recognition performance. A particle swarm is a set of candidate solutions used to optimize model parameters. It iteratively searches for the optimal solution in a search space by simulating particle movement. Each particle represents a set of possible model parameters, and its motion is composed of velocity and position. The particle search strategy is the strategy that controls the movement of the particle swarm in the search space, including position and velocity update rules. In particular, by combining a preset velocity update mechanism with noise interference conditions, the particle swarm can dynamically adapt to different signal environments, improving optimization efficiency.

[0060] Specifically, the computer device performs preprocessing operations on the collected time series signal data set. Preprocessing includes steps such as denoising, data completion, outlier removal, signal alignment and normalization to eliminate interference information in the original signal, and unify the formats of signals under different sampling channels to ensure that all time series signal data are comparable in terms of statistical characteristics and waveform structure. Then, the computer device performs signal enhancement processing on the preprocessed time series signal data set. Preferably, a wavelet threshold method can be used to suppress the noise in the signal through adaptive filtering, while enhancing the amplitude response of the waveform mutation point, so that the signal boundary before and after the fault occurs is clearer, which helps to more accurately identify the fault location.

[0061] Furthermore, to enhance the representation of fault characteristics, computer equipment can combine various processing methods. For example, bandpass filters can be used to retain signals within a known characteristic frequency range while filtering out other frequency components, thereby highlighting the valid signal; signals can be amplified or offset in the time domain to make the characteristics more pronounced; convolution operations can also be used to enhance local features; and signals can be converted to the frequency domain, key frequency components enhanced, and then converted back to the time domain to further remove noise and redundancy.

[0062] The computer then performs feature extraction on the enhanced time series signal dataset. Preferably, a multidimensional feature extraction strategy is employed to extract time-domain features such as the signal's mean, variance, and peak value; frequency-domain features such as the Fourier spectrum and dominant frequency amplitude; and joint time-frequency features such as wavelet energy distribution. Furthermore, features such as the instantaneous mutation amplitude, frequency jump, and harmonic growth rate at the fault starting point are used to characterize the fault's onset and evolution. These features serve as input vectors for the fault identification model.

[0063] Next, the computer device inputs the extracted signal features into a pre-built fault identification model. The fault identification model is trained based on normal signals and fault signals extracted from the cable's historical operation records. Model types include convolutional neural networks, recurrent neural networks, or integrated support vector machines. By initially inputting training data, the computer device establishes an initial particle swarm in the model parameter space, where each particle represents a set of model parameter configurations. The computer device then sets a particle search strategy based on the current noise interference state. This strategy uses a dynamic speed adjustment mechanism. When the signal noise is strong, the particle search step size is increased to escape the local optimum; when the signal is clear, the search speed is reduced to improve the local search accuracy. According to this strategy, the particle position and speed are updated, and it is continuously iterated and optimized.

[0064] Finally, the computer combines the initial particle swarm with the search strategy to perform a search and update operation within a multi-objective optimization framework. In each iteration, the objective function value of each particle is calculated, which comprehensively considers the two optimization aspects of fault location accuracy and response time. During the update process, the particle's historical optimal position and global optimal position are continuously updated. After the particle swarm converges to the optimal region, the parameter configuration is extracted and applied to the fault identification model. The final signal characteristics are input into the model. The computer outputs the cable fault diagnosis result and combines it with information such as the traveling wave propagation time difference, sensor location, and cable topology to accurately generate the cable fault location result.

[0065] After locating the cable fault, the system can assign a warning level based on the fault type, severity, and location, generating a corresponding warning level. Faults of different types and severities carry varying degrees of risk. For example, a short circuit can cause a sudden surge in current, significantly damaging cables and equipment. Faults located in critical areas pose a higher risk. Therefore, based on pre-set criteria, minor, remote faults are assigned a low warning level, while severe, location-sensitive faults are assigned a high warning level. After determining the warning level, the corresponding warning signal is triggered. Low-level warnings may simply display a prompt on the monitoring interface, while medium-level warnings will generate an audible alert. High-level warnings are issued via text, audio, text messages, or push notifications to maintenance personnel. Warnings also include the fault type, severity, and specific location, helping maintenance personnel quickly understand the situation. Warnings can be issued simultaneously via various channels, including large screens, computer terminals, and mobile apps, ensuring that relevant personnel receive timely information and take timely action, minimizing losses and ensuring stable operation of the power system.

[0066] By sequentially performing signal preprocessing, enhancement, feature extraction, and model input, the discriminability of the signal can be effectively improved, redundant information can be compressed, and feature information critical to fault diagnosis can be retained, making the data input to the model highly representative and fault-sensitive. Furthermore, a particle swarm optimization mechanism is introduced to dynamically adjust parameter combinations during model training. Combined with a search strategy based on noise state settings, this guides the particle swarm to rapidly converge on the optimal path in the search space, effectively addressing the limitations of traditional model parameters that rely on manual adjustment. Ultimately, the model achieves higher fault identification accuracy and lower response latency, ensuring rapid and accurate fault location in complex cable operation scenarios.

[0067] In the above embodiment, by rationally configuring distributed sensors according to the identified key positions of the cable, the signal coverage of the key areas is targeted and enhanced, effectively avoiding invalid acquisition and signal loss, and improving the integrity and representativeness of the time series signal data; further, the preprocessing and signal enhancement strategies are adopted to significantly suppress the influence of external electromagnetic interference and noise on the magnetic field traveling wave signal, improve the clarity and feature distinguishability of the signal, and provide a better signal foundation for feature extraction and fault identification; in the fault identification stage, a model parameter adjustment mechanism based on particle swarm optimization is introduced, combined with a particle search strategy considering the noise interference state and an optimization path design of a multi-objective function, so that the model can still maintain high robustness and recognition accuracy under complex working conditions. Therefore, this method effectively enhances the accuracy and practicality of the fault detection system while improving the quality of the cable fault signal.

[0068] In one embodiment, the process of identifying the key locations of the cable includes:

[0069] Extract historical fault information of cables and establish fault hotspot features;

[0070] Obtain the cable load operation task execution results and establish the cable abnormal load characteristics;

[0071] Extract the branching position, cross position and interface position of the cable to construct the cable structural features;

[0072] Determine the critical location of the cable based on the fault hotspot characteristics, cable abnormal load characteristics and cable structure characteristics.

[0073] The cable's historical fault information refers to records of various faults that occurred during the cable's use, typically including the time, location, type, and handling of the fault, reflecting the cable's fault characteristics at different times and under different operating conditions. The cable's load operation task execution results refer to a comprehensive assessment of the cable's operating status obtained by monitoring its electrical parameters during different operation tasks, such as current, voltage, and load curves. The cable's structural characteristics are physical connection point information extracted from cable layout drawings, such as branch points, intersections, and interface locations, reflecting the distribution of the cable's topology and connection complexity.

[0074] Specifically, computer equipment can access cable operation and maintenance logs and online monitoring platforms to extract historical cable fault information. Cluster analysis is performed on data such as fault time, location, and type. The frequency and distribution of faults at different locations are counted, and high-frequency fault areas are identified. Characterization is then performed based on seasonal, ambient temperature, and load variations to generate reusable tags and construct fault hotspot features. Typical fault areas can also be further labeled, such as "prone to overheating" and "high-failure-rate joints," providing a basis for screening key areas.

[0075] The computer performs task extraction, analyzing the cable's real-time and historical electrical parameters, including current, voltage, and power factor, to obtain load data for different time periods. It can identify routine and non-routine tasks (such as emergency power supply) and determine whether abnormal operating conditions exist, such as overload, imbalance, or impact loads. The computer identifies abnormal load characteristics, such as long-term overload, heating risks caused by current asymmetry, or cable damage caused by mechanical shock. The computer annotates the corresponding cable location and anomaly type in the dataset.

[0076] Computer equipment parses structural data from cable design drawings, extracting information about branch points, intersections, and interfaces. Branch points, such as T-junctions, are prone to loops or poor connections; intersections, due to crowded space, can hinder heat dissipation; and interfaces are prone to corrosion or loosening due to the contact of dissimilar materials. Based on structural location and material information, a database of cable structural characteristics is established to further assess structural reliability.

[0077] The computer then fuses these three features, assigns weights to each, and identifies key cable locations based on these rules. For example, a cable location might be classified as high-risk if it simultaneously exhibits frequent historical failures, abnormal current loads, and structural weaknesses. If only two or one of these characteristics are met, it is classified as a medium-risk or low-risk area, respectively. The final output is a list of key location identification results, which can include coordinates, risk level, risk type, and recommended monitoring frequency, enabling visual annotation.

[0078] In this embodiment, fault hotspot features are constructed by extracting historical fault information, helping to identify typical problem areas in cable operation. Abnormal load features are established through load task analysis, reflecting the current operational risks of the cable in real time. Combined with cable structural characteristics, potential risk points can be identified at a physical level. Fusion analysis of these three types of information and the use of multi-dimensional features to determine key locations not only improves positioning accuracy but also enables risk-tiered management of cable faults, effectively enhancing monitoring efficiency and the overall safety and reliability of the cable system.

[0079] In one embodiment, the process of preprocessing the time series signal dataset includes:

[0080] Filter the time series signal data set using a filter;

[0081] Perform interpolation and filling on the filtered time series signal data set;

[0082] Perform signal normalization on the interpolated and padded time series signal dataset.

[0083] Filtering refers to the process of attenuating or removing noise components in a signal that fall outside the target frequency range through a specified filter, thereby enhancing the effective signal. Interpolation and filling refers to the process of estimating and supplementing missing or abnormal data based on existing data points when there are missing values ​​in the signal or when sampling points are unevenly spaced, thereby maintaining signal continuity along the time axis. Normalization refers to the process of standardizing signal data from different sources or units to a uniform numerical range.

[0084] Specifically, the computer device can first filter the collected time-series signal dataset. Specifically, based on the spectral distribution characteristics of the cable magnetic field traveling wave signal, an appropriate filter type can be selected for processing. For example, if the target signal is primarily concentrated in the low-frequency band and there is high-frequency interference, a low-pass filter can be used. If it is necessary to retain the signal within a certain frequency range while suppressing noise in the frequency bands on both sides, a band-pass filter can be used. This filtering rule is applied to the entire dataset, performing frequency domain filtering operations on each sampling point, resulting in a noise-reduced filtering result.

[0085] The computer checks the filtering results for missing data or uneven sampling. If there are discontinuities between sampling points or missing data for certain time periods, the computer activates the interpolation module, using linear or high-order polynomial interpolation methods to calculate and fill in appropriate data points in the missing intervals, thereby restoring the temporal continuity of the signal. This ensures a stable sampling rate and data integrity for the model input signal, preventing information discontinuities from affecting the recognition effect.

[0086] After interpolation is complete, the computer performs normalization on the interpolated time series signal dataset. The computer calculates the minimum and maximum values ​​for each signal channel and maps the original signal values ​​to a uniform numerical range using a normalization formula. If multiple sensors are simultaneously acquiring signals, all channels are normalized separately to eliminate the effects of different acquisition channels, sensor devices, or measurement units, thereby ensuring that the subsequent fault identification model can uniformly process signals from different sources.

[0087] In this implementation, filtering effectively removes non-target frequency noise components from the cable magnetic field traveling wave signal, significantly improving signal quality and enhancing effective feature information. Interpolation and filling address data gaps caused by noise processing or uneven sampling, ensuring signal continuity and stability, and helping to maintain the characteristic time series structure. Furthermore, standardization unifies signals collected by different sensors to the same dimension and numerical range, improving signal comparability and model recognition consistency. This improves the adaptability and accuracy of the cable fault identification model for multi-source heterogeneous signals.

[0088] In one embodiment, the optimization process of the fault identification model includes:

[0089] Establishing a particle swarm in an initial state according to the initial model parameters of the fault identification model, and determining the current position and current speed of each particle in the particle swarm;

[0090] In each round of iteration, after determining the current noise interference state corresponding to the current particle swarm, the inertia weight of the current particle swarm is adjusted, and the current position and current velocity of each particle in the current particle swarm are updated. The fitness of the updated particle swarm is evaluated according to the multi-objective function until the fitness evaluation result of the updated particle swarm meets the preset update end condition, and the optimal model parameters of the fault recognition model are obtained.

[0091] The particle velocity represents the particle's movement trend and step size in the parameter space. The inertia weight controls the particle's ability to maintain its current direction of motion during the search process, affecting the search range and convergence speed. The noise interference state describes the degree to which external factors such as electromagnetic interference, humidity changes, and power frequency interference affect signal quality during signal acquisition.

[0092] Specifically, based on the parameter structure of the fault identification model, the computer device first defines the parameter dimensions and value ranges required for model optimization, such as weight parameters in the neural network structure, threshold settings, or sensitivity factors in the feature extraction module. It then generates an initial set of individual particles, sets the total number of particles N (e.g., N = 50), and constructs a parameter vector for each particle in the D-dimensional parameter space. The particle's position vector represents a set of feasible model parameter combinations, while the velocity vector represents the particle's movement direction and step size in the parameter space.

[0093] Subsequently, in each round of iteration, the computer equipment first collects the current fault detection signal data and identifies the current noise interference state based on the interference scoring function. For example, when the humidity in the detection signal suddenly increases and the power frequency noise accounts for more than 50%, it is judged that the current state is a strong interference state. According to the judgment result, the inertia weight of the particle swarm is dynamically adjusted. If the current environment is stable and the interference is low, the inertia weight value is increased, such as 0.9, to encourage the particles to perform global exploration at a higher speed; if the interference level is moderate, a linear weight decay strategy is implemented, such as from 0.7 to 0.5, to achieve a balance between convergence and exploration; if it is a strong interference state, a smaller inertia weight is set, such as 0.3, to encourage particles to perform local fine search with small steps.

[0094] Next, based on the updated inertia weights, the particle's position and velocity are updated. This means the particle's next movement path is calculated based on factors such as its current velocity, the individual's historical optimal position, and the group's optimal position, and its parameter combination is updated. After each update, a multi-objective function is used to evaluate the fitness of each particle's solution. The optimal solution for the current round is selected by comprehensively considering metrics such as the model's recognition accuracy, error rate, and operational efficiency.

[0095] Finally, when the overall fitness of the particle swarm reaches the preset termination condition during the iteration process, such as no significant change in the optimal fitness for several consecutive rounds or the maximum number of iterations is reached, the computer device terminates the particle swarm optimization process and outputs the model parameters corresponding to the particle with the highest current fitness as the optimal model parameters of the fault identification model, realizing accurate parameter configuration of the model under complex working conditions.

[0096] In this embodiment, by establishing a particle swarm and dynamically adjusting its inertia weights, an adaptive search strategy for fault identification model parameters can be implemented for varying noise interference environments. Introducing a noise-state-based weight adjustment mechanism enhances the particle swarm optimization algorithm's sensitivity to changes in signal quality, thereby improving the model's adaptability to dynamic environmental changes. Ultimately, the fault identification model maintains high accuracy and robustness even in complex operating environments, enabling stable identification of fault points and optimal parameter configuration.

[0097] In one embodiment, the step of determining the current noise interference state corresponding to the current particle swarm includes:

[0098] Establishing an external environment feature set based on the collected external environment data of the cable;

[0099] Perform multi-dimensional noise quantification on the external environment feature set to obtain the external noise interference score;

[0100] Perform noise anomaly evaluation on the signal characteristics corresponding to each particle in the current particle swarm to obtain the current signal noise interference score;

[0101] The external noise interference score and the current signal noise interference score are used as the current noise interference state.

[0102] The external environment feature set refers to a multidimensional parameter set that characterizes the external operating conditions of the cable, constructed by computer equipment based on various environmental information data collected at the cable operation site, after feature extraction and analysis. This parameter includes, but is not limited to, electromagnetic interference frequency, relative humidity, vibration intensity, temperature gradient, and other dimensions, reflecting the impact of the cable environment on the stability of the electromagnetic signal. The external noise interference score refers to the comprehensive score obtained by the computer equipment based on this environmental feature set, after mapping each type of interference into a quantitative score through quantitative rules, and performing a weighted summation of different interference types to assess the intensity of interference from the external environment on the cable signal. The signal noise interference score refers to the score value obtained after the noise anomaly evaluation of the cable signal characteristics corresponding to the current particle is performed, reflecting the degree of interference pollution to the signal itself.

[0103] Specifically, after receiving data from multiple types of environmental sensors deployed on-site, the computer equipment first extracts multidimensional features of the external environment and establishes an external environment feature set. Specifically, the computer equipment connects to an electromagnetic radiation detector, spectrum analyzer, temperature and humidity sensors, accelerometers, and other sensors to continuously monitor the operating environment of the cable area. The computer equipment performs a Fourier transform on the acquired raw signal data to extract the dominant frequency components of electromagnetic and vibration interference. It also performs statistical analysis on time series data such as temperature and humidity, extracting statistics such as mean, extreme values, variance, and temporal trends. Combined with historical cable fault data, the computer equipment establishes correlations between external environmental features and fault signals, ultimately forming a multidimensional external environment feature set.

[0104] Subsequently, the computer device performs a multi-dimensional noise quantification identification operation based on the feature set and in accordance with preset quantization rules to obtain an external noise interference score. Preferably, the computer device scores electromagnetic interference, humidity interference, vibration interference, and temperature interference separately according to the interference type quantization data mapping table shown in the following table, and performs a weighted summation operation on each score result according to the set weight, with the score range being 0 to 100 points, thereby forming a unified environmental noise interference intensity index.

[0105]

[0106] Next, the signal features corresponding to the current particle are extracted to identify noise anomalies and establish a second noise interference score. Specifically, the computer device uses the model input feature vector represented by the current particle as the analysis object, and extracts the key feature dimensions related to noise in the vector, including the baseline drift amplitude and pulse noise frequency in the time domain; the energy ratio of the power frequency component in the frequency domain; and the distribution position of sudden interference in the time-frequency diagram. Based on a historical sample library with annotated noise levels, the computer device calls a pre-trained classification model to map the signal to the noise interference level, and uses the classification probability output value as the scoring basis to finally generate a signal-noise interference score.

[0107] Finally, the computer device inputs the external noise interference score and the signal noise interference score as the current noise interference state into the particle swarm control module. Based on the scoring results, the computer device dynamically sets the search strategy of the particle swarm optimization algorithm. For example, if both scores are low, the particle swarm is set to a low interference state, and the inertia weight is increased to enhance global search capabilities. If the scores are medium, the inertia weight is gradually linearly decayed. If both scores are high or an abnormal feature is detected, such as a sudden increase in humidity or strong electromagnetic interference, it is determined to be a strong interference state. The particle inertia weight is fixed to a small value, prompting the particle swarm to enter a local fine search state to ensure search accuracy.

[0108] In this embodiment, by constructing an external environment feature set and performing multi-dimensional noise identification on it, not only can the interference intensity of the cable operating environment be fully reflected, but the external environmental factors can also be quantified into interference state parameters that can be identified by the optimization algorithm; at the same time, noise anomaly identification is performed on the signal characteristics corresponding to the current particle, which can truly reflect the pollution level of the signal itself. The external noise interference score and the signal noise interference score are combined to form the current noise interference state, which can achieve adaptive modeling control under different working conditions. By adjusting the inertia weight parameter in the particle swarm optimization strategy, the particles can perform local fine search in high-interference environments and global rapid exploration in low-interference environments, which is conducive to enhancing the stability and accuracy of fault identification model parameter optimization, thereby significantly improving the applicability and accuracy of cable fault detection in complex environments.

[0109] In one embodiment, the step of updating the current position and current velocity of each particle in the current particle swarm includes:

[0110] According to the number of iterations of the current particle swarm, the current benchmark iteration speed is established;

[0111] Determine the current tracking update speed based on the previous round position and historical optimal position of each particle in the current particle swarm, as well as the global optimal position of the current particle swarm;

[0112] The current position and current velocity of each particle in the current particle swarm are determined according to the current benchmark iteration velocity, the current tracking update velocity, and the adjusted inertia weight.

[0113] The baseline iteration speed refers to the basic movement speed parameter set by the computer based on the number of iterations of the current particle swarm and the stage characteristics of the search process. It determines the basic search range of the particle in the parameter space. The tracking update speed refers to the direction and speed calculated by the particle based on the relative relationship between its position in the previous round, its historical optimal position, and the current global optimal position of the particle swarm. It reflects the particle's intention to move towards a better solution.

[0114] Specifically, the computer device obtains the current number of particle swarm iterations and sets the current baseline iteration speed accordingly. Preferably, in the early stages of the particle swarm optimization algorithm, such as when the number of iterations is below a preset threshold, the computer device sets a relatively high baseline speed to enhance the particles' ability to jump in the parameter space, thereby expanding the search range and quickly discovering potential solution regions. In the middle and late stages of the iterations, the computer device gradually reduces the baseline iteration speed to a lower value by performing nonlinear or linear attenuation on the number of iterations, guiding the particles to gradually converge to more detailed regions and improving local search accuracy.

[0115] Next, the computer device obtains the previous round position, historical optimal position, and current global optimal position of each particle in the current particle swarm. Based on the vector distance and direction relationship between the three positions, the computer device performs a tracking update analysis on each particle. First, the offset vector between the particle's current direction and its historical optimal position is calculated, and then the offset vector relative to the global optimal position is calculated. The two vectors are then superimposed at a certain ratio to obtain the current tracking update speed. This speed is used to describe the particle's adaptive approach behavior under the dual influence of local experience and group guidance, reflecting the particle's intelligent search strategy of "learning" and "collaboration."

[0116] After obtaining the baseline iteration speed and the tracking update speed, the computer integrates them to create a more adaptable speed update mechanism. In actual execution, the computer considers each particle's current speed, its best historical position, its current position, and the global optimal position found by the entire particle swarm over previous iterations. Based on this, the computer adjusts the particle's movement direction and speed according to pre-set update rules, combining the particle's own historical experience with guidance from the swarm's optimal solution.

[0117] While adjusting the speed, the computer also updates the particle's current position based on the updated speed information. That is, the particle moves from its old position to its new position based on its current movement trend, completing the iteration process of this round of parameter search. By continuously updating speed and position, the particle can gradually approach a more optimal solution within the parameter space, improving search efficiency and accuracy. Furthermore, to further enhance the search's adaptability, the computer also dynamically adjusts the inertia weight parameter. The inertia weight controls the degree to which the particle retains its previous velocity. In the initial stages of the algorithm, the computer sets a large inertia weight value to maintain a strong global search capability, facilitating extensive exploration of the parameter space. As the iterations proceed, this weight value is gradually reduced, making the particle's search more refined and facilitating local searches near the optimal solution, thereby improving the accuracy of the final convergence.

[0118] In this embodiment, setting a baseline iteration speed during particle swarm optimization helps control the overall pace of the search and achieve a dynamic transition from rough exploration to refined search. The introduction of a follow-up update speed allows each particle to be driven not only by its own experience but also by the collaborative progress of the entire swarm, thereby improving the convergence quality of the search. The fusion calculation of these two speeds, combined with the dynamic adjustment of the inertia weight, gives the particle swarm greater search flexibility and precision adaptability throughout the optimization process. This can significantly improve the efficiency of fault identification model parameter optimization and enhance the model's recognition robustness and generalization capabilities in complex cable environments.

[0119] In one embodiment, the process of establishing the multi-objective function includes:

[0120] Analyze cable historical fault data and cable normal operation data to extract cable signal characteristics corresponding to different fault stages;

[0121] According to the cable signal characteristics corresponding to each fault stage, initial weights are assigned to various evaluation targets, including positioning accuracy target, signal accuracy target, calculation speed target, and real-time target.

[0122] A time constraint for adjusting weights over time is constructed, and the weights of various evaluation objectives are adjusted to obtain a multi-objective function.

[0123] Historical cable fault data refers to signal data, location information, and operating parameters related to cable faults collected or received by computer equipment. This data comes from sources such as historical case libraries and on-site monitoring systems. Normal cable operation data refers to various operating parameters and corresponding signals when the cable is in a fault-free state, serving as a reference for fault identification. Cable signal characteristics represent key indicators used to characterize the cable's operating status, such as amplitude, frequency, and discharge intensity. Fuzzy fault stages refer to different fault development stages, including early hidden danger stage, development stage, and impending fault stage, based on the changing trends of signal characteristics. Evaluation objectives, including positioning accuracy, signal accuracy, computation speed, and real-time performance, are key indicators for measuring the performance of cable fault identification models. The positioning accuracy objective refers to the degree of closeness between the fault location result and the actual fault point. The smaller the error, the higher the positioning accuracy. The signal accuracy objective refers to the consistency between the processed cable signal and the original real signal. The higher the consistency, the better the signal accuracy. The computation speed objective refers to the processing time required from signal reception to result output. The shorter this time, the faster the computation speed. The real-time performance objective refers to the system's ability to complete fault identification and location within a specified timeframe. The more timely the response, the higher the real-time performance. Time constraint is a strategic condition for dynamically adjusting the weights of each evaluation objective as the fault develops.

[0124] Specifically, the computer device reads the cable's historical fault data and normal operating data, normalizes and denoises them, and then performs feature extraction. The computer device can identify the characteristic distribution of cable signals under different operating conditions based on machine learning algorithms or statistical analysis methods, thereby associating these characteristics with known fault types. Next, the computer device divides the cable fault process into stages to obtain multiple different fault stages. For example, the computer device identifies a state where the partial discharge signal is weak and the insulation resistance decreases slightly as an early hidden danger stage, and identifies a state where the signal frequency increases abnormally or the insulation resistance decreases rapidly as a development stage or an impending fault stage. The computer device analyzes the system response requirements corresponding to each fault stage and determines the degree of dependence on each evaluation target.

[0125] The computer then assigns initial weights to the four evaluation objectives: positioning accuracy, signal accuracy, calculation speed, and real-time performance. For example, in the early stages, signal accuracy is given the highest weight to more effectively identify subtle signal changes. However, in the near-fault phase, the weights for real-time performance and positioning accuracy are significantly increased to ensure a rapid response. Weights can be determined through empirical rules, expert systems, or models trained on data.

[0126] Furthermore, the computer equipment incorporates a time constraint mechanism, which dynamically adjusts the weights of the aforementioned evaluation objectives based on fault development time or phase transition events. For example, during operation, if a cable is determined to have entered the development phase from the early stage, the weight of the signal accuracy objective will be automatically lowered by a preset ratio, while the weights of real-time performance and computational speed will be increased. The multi-objective function is then restructured to ensure that subsequent optimization is more aligned with the current phase's requirements.

[0127] Finally, the computer applies the dynamically updated multi-objective function to the fitness calculation within the particle swarm optimization algorithm. Based on how well the model parameter combination represented by the current particle performs under this function, the fitness of the particle is evaluated, guiding the particle swarm's next search path. This process iterates continuously to find the optimal parameters for the fault identification model, ultimately improving fault location accuracy and response efficiency.

[0128] In this embodiment, because the cable has different emphases on model performance at different fault stages, constructing a multi-objective function using only fixed weights can easily lead to performance degradation at a particular stage. Therefore, by introducing fault stage division, the computer device can determine the stage based on the current signal state, thereby rationally adjusting the weights of each objective in the multi-objective function. Combining time constraints with weight adjustment strategies not only improves the model's adaptability in different stages, but also guides the particle swarm algorithm to optimize in a direction that meets actual needs within different time periods, significantly enhancing the overall performance of fault identification and location. This balances accuracy and efficiency, providing stable and reliable technical support for online fault detection and response in complex cable operation scenarios.

[0129] The following describes the magnetic field traveling wave signal processing device for cable fault detection provided by the embodiment of the present application. The magnetic field traveling wave signal processing device for cable fault detection described below and the magnetic field traveling wave signal processing method for cable fault detection described above can be referred to each other. Figure 2 As shown, the present application provides a magnetic field traveling wave signal processing device for cable fault detection, the device comprising:

[0130] The time series signal data set determination module 201 is used to configure distributed sensors according to the identified key positions of the cable, and collect the cable's magnetic field traveling wave signal through the distributed sensors to obtain a time series signal data set;

[0131] The fault location result determination module 202 is used to preprocess and enhance the time series signal data set in sequence, and input the signal characteristics of the enhanced time series signal data set into a pre-optimized fault identification model to obtain the cable fault location result;

[0132] Among them, the fault identification model is based on a particle swarm and a preset particle search strategy, and is obtained by optimizing the model parameters using a pre-established multi-objective function. The particle search strategy is set according to a preset speed update mechanism and the current noise interference state.

[0133] In one embodiment, the timing signal data set determining module 201 includes:

[0134] Fault hotspot feature establishment unit, used to extract historical fault information of the cable and establish fault hotspot features;

[0135] A cable abnormal load feature establishment unit is used to obtain the cable load operation task execution result and establish the cable abnormal load feature;

[0136] The cable structure feature construction unit is used to extract the branch position, cross position and interface position of the cable and construct the cable structure feature;

[0137] The key position determination unit is used to determine the key position of the cable according to the fault hot spot characteristics, cable abnormal load characteristics and cable structure characteristics.

[0138] In one embodiment, the fault location result determination module 202 includes:

[0139] A time series signal data set filtering processing unit, configured to filter the time series signal data set using a filter;

[0140] A time series signal data set interpolation and filling unit, used to interpolate and fill the time series signal data set after filtering;

[0141] The time series signal data set signal standardization processing unit is used to perform signal standardization processing on the time series signal data set after interpolation and filling.

[0142] In one embodiment, the fault location result determination module 202 includes:

[0143] A particle swarm establishment unit is used to establish a particle swarm in an initial state according to the initial model parameters of the fault identification model, and to determine the current position and current speed of each particle in the particle swarm;

[0144] The fault identification model optimization unit is used to determine the current noise interference state corresponding to the current particle swarm in each round of iteration, adjust the inertia weight of the current particle swarm, and update the current position and current velocity of each particle in the current particle swarm. The fitness of the updated particle swarm is evaluated according to the multi-objective function until the fitness evaluation result of the updated particle swarm meets the preset update end condition, thereby obtaining the optimal model parameters of the fault identification model.

[0145] In one embodiment, the fault identification model optimization unit includes:

[0146] The external environment feature set establishing subunit is used to establish the external environment feature set based on the collected external environment data of the cable;

[0147] The external noise interference score determination subunit is used to perform multi-dimensional noise quantitative identification on the external environment feature set to obtain the external noise interference score;

[0148] The current signal-noise interference score determination subunit is used to perform noise anomaly evaluation on the signal characteristics corresponding to each particle in the current particle swarm to obtain the current signal-noise interference score;

[0149] The current noise interference state determining subunit is configured to use the external noise interference score and the current signal noise interference score as the current noise interference state.

[0150] In one embodiment, the fault identification model optimization unit includes:

[0151] The current benchmark iteration speed establishment subunit is used to establish the current benchmark iteration speed according to the number of iteration rounds of the current particle swarm;

[0152] The current following update speed determination subunit is used to determine the current following update speed according to the previous round position and historical optimal position of each particle in the current particle swarm, as well as the global optimal position of the current particle swarm;

[0153] The particle swarm update subunit is used to determine the current position and current speed of each particle in the current particle swarm according to the current benchmark iteration speed, the current tracking update speed and the adjusted inertia weight.

[0154] In one embodiment, the fault location result determination module 202 includes:

[0155] The cable signal feature extraction unit is used to analyze the cable's historical fault data and the cable's normal operation data, and extract the cable signal features corresponding to different fault stages;

[0156] An initial weight allocation unit is used to allocate initial weights to various evaluation targets based on the cable signal characteristics corresponding to each fault stage, wherein the evaluation targets include positioning accuracy target, signal accuracy target, calculation speed target, and real-time target;

[0157] The multi-objective function determination unit is used to construct a time constraint for adjusting the weight over time, adjust the weight of each evaluation target, and obtain a multi-objective function.

[0158] In one embodiment, the present application also provides a storage medium storing computer-readable instructions. When the computer-readable instructions are executed by one or more processors, the one or more processors execute the steps of the magnetic field traveling wave signal processing method for cable fault detection as described in any of the above embodiments.

[0159] In one embodiment, the present application also provides a computer device having computer-readable instructions stored therein. When the computer-readable instructions are executed by one or more processors, the one or more processors execute the steps of the magnetic field traveling wave signal processing method for cable fault detection as described in any of the above embodiments.

[0160] Schematically, as Figure 3 As shown, Figure 3 This is a schematic diagram of the internal structure of a computer device provided in an embodiment of the present application. The computer device 300 can be provided as a server. Figure 3Computer device 300 includes a processing component 302, which further includes one or more processors, and memory resources represented by memory 301 for storing instructions executable by processing component 302, such as application programs. The application programs stored in memory 301 may include one or more modules, each corresponding to a set of instructions. Furthermore, processing component 302 is configured to execute the instructions to perform the method for processing traveling wave magnetic field signals for cable fault detection according to any of the aforementioned embodiments.

[0161] The computer device 300 may further include a power supply component 303 configured to perform power management of the computer device 300, a wired or wireless network interface 304 configured to connect the computer device 300 to a network, and an input / output (I / O) interface 305. The computer device 300 may operate based on an operating system stored in the memory 301, such as Windows Server™, Mac OS X™, Unix™, Linux™, Free BSD™, or the like.

[0162] Those skilled in the art will understand that Figure 3 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0163] Finally, it should be noted that, in this article, relational terms such as first and second are merely used to distinguish one entity or operation from another, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprise," "include," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, article, or device comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or elements inherent to such process, method, article, or device. Without further restriction, an element defined by the phrase "comprising a..." does not exclude the presence of other identical elements in the process, method, article, or device comprising the element. Herein, "one," "said," "the," and "its" may also include plural forms unless the context clearly indicates otherwise. A plurality refers to at least two, such as 2, 3, 5, or 8. "And / or" includes any and all combinations of the relevant listed items.

[0164] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The various embodiments can be combined as needed, and the same or similar parts can be referenced to each other.

[0165] The above description of the disclosed embodiments is intended to enable one skilled in the art to implement or use the present application. Various modifications to these embodiments will be readily apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application is not limited to the embodiments shown herein, but is intended to conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for processing magnetic field traveling wave signals for cable fault detection, characterized in that: The method comprises: Distributed sensors are configured according to the identified key positions of the cable, and the distributed sensors are used to collect the traveling wave signals of the magnetic field of the cable to obtain a time series signal data set; Preprocessing and signal enhancement are performed on the time series signal data set in sequence, and signal features of the signal-enhanced time series signal data set are input into a pre-optimized fault identification model to obtain a fault location result of the cable; The fault identification model is obtained by optimizing model parameters based on a particle swarm and a preset particle search strategy using a pre-established multi-objective function. The particle search strategy is set according to a preset speed update mechanism and the current noise interference state.

2. The method for processing magnetic field traveling wave signals for cable fault detection according to claim 1, characterized in that: The process of identifying the key positions of the cable includes: Extracting historical fault information of the cable and establishing fault hotspot features; Obtaining the load operation task execution result of the cable and establishing abnormal load characteristics of the cable; Extracting the branch position, cross position and interface position of the cable to construct cable structural features; The critical position of the cable is determined according to the fault hot spot characteristics, the cable abnormal load characteristics and the cable structure characteristics.

3. The method for processing magnetic field traveling wave signals for cable fault detection according to claim 1, characterized in that: The process of preprocessing the time series signal data set includes: Performing filtering processing on the time series signal data set using a filter; Perform interpolation and filling on the filtered time series signal data set; Perform signal normalization on the interpolated and padded time series signal dataset.

4. The method for processing magnetic field traveling wave signals for cable fault detection according to claim 1, characterized in that: The optimization process of the fault identification model includes: Establishing a particle swarm in an initial state according to the initial model parameters of the fault identification model, and determining the current position and current speed of each particle in the particle swarm; In each round of iteration, after determining the current noise interference state corresponding to the current particle swarm, the inertia weight of the current particle swarm is adjusted, and the current position and current velocity of each particle in the current particle swarm are updated. The fitness of the updated particle swarm is evaluated according to the multi-objective function until the fitness evaluation result of the updated particle swarm meets the preset update end condition, thereby obtaining the optimal model parameters of the fault identification model.

5. The method for processing magnetic field traveling wave signals for cable fault detection according to claim 4, characterized in that: The step of determining the current noise interference state corresponding to the current particle swarm includes: Establishing an external environment feature set based on the collected external environment data of the cable; Performing multi-dimensional noise quantification identification on the external environment feature set to obtain an external noise interference score; Perform noise anomaly evaluation on the signal characteristics corresponding to each particle in the current particle swarm to obtain the current signal noise interference score; The external noise interference score and the current signal noise interference score are used as the current noise interference state.

6. The method for processing magnetic field traveling wave signals for cable fault detection according to claim 4, characterized in that: The step of updating the current position and current velocity of each particle in the current particle swarm includes: According to the number of iterations of the current particle swarm, the current benchmark iteration speed is established; Determine the current tracking update speed based on the previous round position and historical optimal position of each particle in the current particle swarm, as well as the global optimal position of the current particle swarm; The current position and current velocity of each particle in the current particle swarm are determined according to the current benchmark iteration velocity, the current tracking update velocity, and the adjusted inertia weight.

7. The method for processing magnetic field traveling wave signals for cable fault detection according to any one of claims 1 to 6, characterized in that: The process of establishing the multi-objective function includes: Analyze cable historical fault data and cable normal operation data to extract cable signal characteristics corresponding to different fault stages; According to the cable signal characteristics corresponding to each fault stage, initial weights are assigned to the evaluation targets, wherein the evaluation targets include positioning accuracy target, signal accuracy target, calculation speed target, and real-time target; A time constraint for adjusting the weight over time is constructed, and the weight of each evaluation target is adjusted to obtain the multi-objective function.

8. A magnetic field traveling wave signal processing device for cable fault detection, characterized in that: The device comprises: A time series signal data set determination module is used to configure distributed sensors according to the identified key positions of the cable, and collect the magnetic field traveling wave signal of the cable through the distributed sensors to obtain a time series signal data set; a fault location result determination module, configured to sequentially preprocess and enhance the time series signal data set, and input the signal characteristics of the enhanced time series signal data set into a pre-optimized fault identification model to obtain a fault location result of the cable; The fault identification model is obtained by optimizing model parameters based on a particle swarm and a preset particle search strategy using a pre-established multi-objective function. The particle search strategy is set according to a preset speed update mechanism and the current noise interference state.

9. A storage medium, characterized in that: The storage medium stores computer-readable instructions, which, when executed by one or more processors, enable the one or more processors to perform the steps of the magnetic field traveling wave signal processing method for cable fault detection as claimed in any one of claims 1 to 7.

10. A computer device, characterized in that: include: one or more processors, and memory; The memory stores computer-readable instructions, which, when executed by the one or more processors, execute the steps of the method for processing magnetic field traveling wave signals for cable fault detection according to any one of claims 1 to 7.

Images