Multi-scene adaptive electrical circuit fault positioning and automatic repairing system

Abstract

The invention relates to a multi-scene adaptive electrical circuit fault positioning and automatic repairing system, which comprises the following steps of: constructing a unified multi-modal fault data cube through time-space alignment and fusion processing of multi-source heterogeneous data; electrical characteristics and physical representation characteristics are analyzed in parallel by adopting a double-flow collaborative feature extraction architecture, and accurate three-dimensional positioning of a fault point and depth representation of fault essence are realized; calculating a restoration complexity pre-judgment index based on multi-factor fusion, and comprehensively considering fault features, environmental constraints and real-time meteorological conditions; dynamically matching repair resources and planning an optimal operation path through a self-adaptive repair strategy generation mechanism, and ensuring that a repair scheme is matched with a field working condition; and establishing a closed-loop optimization system based on real-time feedback, and continuously adjusting decision parameters through a continuous learning mechanism. According to the method, the accuracy and efficiency of fault processing can be effectively improved, the adaptability and reliability in multiple scenes are enhanced, and technical support is provided for realizing intelligentization and autonomy of intelligent power grid fault processing.

Description

Technical Field

[0001] This invention relates to the fields of artificial intelligence and power system operation and maintenance technology, specifically to a multi-scenario adaptable electrical line fault location and automatic repair system. Background Technology

[0002] Fault location and automatic repair methods for electrical lines are key technologies in the operation and maintenance of modern smart grids. These methods rapidly pinpoint fault locations and automatically execute repair operations after a fault occurs, minimizing power outage time and improving power supply reliability. Typically, these methods require the comprehensive utilization of various monitoring data along the line, combined with intelligent algorithms to achieve accurate fault diagnosis and repair decisions, adapting to complex and diverse line operating environments.

[0003] However, existing fault handling technologies still have significant limitations when dealing with multi-scenario applications. On the one hand, data collected by different monitoring systems are often independent, making it difficult to form a comprehensive understanding of the fault situation, resulting in inaccurate location results. On the other hand, traditional repair strategies mainly rely on preset rules, failing to fully consider actual factors such as the physical damage at the fault site and environmental conditions, making the repair solutions lack specificity. More importantly, existing systems generally lack self-optimization capabilities and cannot dynamically adjust strategies based on actual repair results, leading to poor handling performance when facing new fault scenarios. This situation of insufficient data utilization, rigid decision-making processes, and insufficient system adaptability severely restricts further improvements in fault handling efficiency. Summary of the Invention

[0004] Based on this, the purpose of the present invention is to provide an electrical circuit fault location and automatic repair system that can comprehensively perceive fault information, accurately predict repair difficulty, and has self-learning capabilities and adapt to multiple scenarios.

[0005] The objective of this invention is achieved through the following solution:

[0006] This invention provides a multi-scenario adaptable electrical circuit fault location and automatic repair system, which is configured with the following modules:

[0007] The multimodal spatiotemporal fusion module is used to perform spatiotemporal alignment and fusion processing on electrical and visual data from different timing synchronization sensing devices, unifying fault information from different sources into the same spatiotemporal coordinate system and generating a standardized multimodal fault data cube.

[0008] The dual-stream collaborative feature extraction module is used to perform dual-stream collaborative feature extraction on the multimodal fault data cube. It extracts the electrical characteristics and physical appearance characteristics of the fault through the parallel processing electrical feature extraction branch and visual feature extraction branch, respectively, and generates the accurate three-dimensional coordinates of the fault point and the in-depth fault feature vector.

[0009] The repair complexity prediction module is used to acquire real-time geographic environmental data of the area where the fault point is located, and calculate the repair complexity prediction index by combining the deepened fault feature vector and the precise three-dimensional coordinates. The repair complexity prediction index value is generated by weighting and nonlinearly transforming the deepened fault feature vector with environmental accessibility parameters and real-time meteorological parameters.

[0010] The adaptive repair strategy generation module is used to generate adaptive repair strategies based on the predicted repair complexity index and the in-depth fault feature vector. It matches repair resources according to the preset resource-capability mapping relationship and plans the operation path to the fault point. It generates a personalized repair work order containing the execution unit, target location and operation procedure and sends it to the repair unit to be dispatched.

[0011] The repair strategy optimization module is used to optimize the strategy based on the actual operation data stream fed back by the repair unit. By comparing the difference between the actual operation data and the expected parameters of the work order, it uses the online learning capability of the feature extraction network to adjust the weight parameters of the resource-capability mapping relationship and generate the optimized resource-capability mapping relationship.

[0012] In one embodiment, the multimodal spatiotemporal fusion module of the electrical line fault location and automatic repair system provided by the present invention is configured with the following units:

[0013] The electrical signal preprocessing unit is used to preprocess the raw electrical signals collected by the traveling wave sensor group and the fault recorder, extract effective fault feature waveforms through filtering and noise cancellation techniques, and generate high-frequency transient current traveling wave sequences and power frequency voltage and current drop waveforms.

[0014] The image data quality enhancement unit is used to enhance the quality of the raw image data collected by inspection drones and fixed monitoring devices. It uses image deblurring and contrast enhancement algorithms to improve the visual effect and generate visible light image sequences and infrared thermal imaging sequences.

[0015] The spatiotemporal synchronization mapping processing unit is used to perform time synchronization processing on high-frequency transient current traveling wave sequences and power frequency voltage and current drop waveforms, unify the time reference of all electrical signals and extract key waveform features. At the same time, it performs spatial coordinate mapping processing on visible light image sequences and infrared thermal imaging sequences, accurately associates visual data with the line geographic information system, and generates a multimodal fault data cube.

[0016] In one embodiment, the dual-stream collaborative feature extraction module of the electrical line fault location and automatic repair system provided by the present invention is configured with the following units:

[0017] The electrical multi-scale feature extraction unit is used to perform multi-scale feature extraction processing on electrical quantity data in the multimodal fault data cube. It mines fault features in different frequency bands through wavelet packet transform and deep convolutional network to generate electrical feature vectors containing transient and steady-state characteristics.

[0018] The visual attention enhancement unit is used to enhance the visual data in the multimodal fault data cube through an attention mechanism. It uses the spatial attention module to focus on the visual anomaly features of the fault area and generate a visual feature vector with spatial positioning capability.

[0019] The cross-modal feature fusion unit is used to perform cross-modal feature fusion processing on electrical feature vectors and visual feature vectors. Through feature splicing and fully connected layers, it learns the correlation between modes and generates a comprehensive fault feature vector and accurate three-dimensional coordinates that reflect the fault characteristics.

[0020] In one embodiment, the repair complexity prediction module of the electrical circuit fault location and automatic repair system provided by the present invention is configured with the following units:

[0021] The environmental accessibility assessment unit is used to acquire geographic environmental data of the area where the fault point is located in real time, process the geographic environmental data of the area where the fault point is located in real time, obtain terrain slope, vegetation density and traffic accessibility parameters by querying the geographic information system, and generate an environmental accessibility assessment vector.

[0022] The meteorological impact assessment unit is used to collect and process meteorological monitoring data in the area where the fault point is located in real time. It obtains real-time parameters such as wind speed, precipitation and temperature by connecting to the meteorological data interface and generates a meteorological impact assessment vector.

[0023] The repair complexity prediction unit is used to perform multi-factor fusion calculations on the in-depth fault feature vector, environmental accessibility assessment vector, and meteorological impact assessment vector. It evaluates the difficulty of repair operations through weighted summation and nonlinear activation function, and generates a repair complexity prediction index value.

[0024] In one embodiment, the adaptive repair strategy generation module of the electrical line fault location and automatic repair system provided by the present invention is configured with the following units:

[0025] The repair resource matching unit is used to intelligently match repair resources based on the repair complexity prediction index value. It selects suitable repair equipment combinations from the resource library according to a preset threshold range and generates the optimal repair unit set.

[0026] The three-dimensional path planning unit is used to perform three-dimensional path planning based on the pre-set resource-capacity mapping relationship and the deepened fault feature vector, considering the fault location and environmental accessibility parameters. It comprehensively considers terrain obstacle avoidance and equipment operation capability limitations to calculate the optimal travel trajectory and generate a safe and efficient operation path scheme.

[0027] The repair work order synthesis unit is used to automatically synthesize work orders from repair unit sets and work path plans. It combines fault type and repair complexity to generate detailed operation instructions and emergency plans, and generates executable personalized repair work orders.

[0028] In one embodiment, the repair strategy optimization module of the electrical line fault location and automatic repair system provided by the present invention is configured with the following units:

[0029] The operation status monitoring unit is used to monitor and process the operation status data fed back by the repair unit in real time. It collects information such as the robot arm posture, operation progress and environmental changes through sensors to generate real-time operation data stream.

[0030] The digital twin simulation unit is used to perform digital twin simulation based on the deepened fault feature vector and environmental accessibility assessment vector. It predicts the operation parameters of each stage by establishing a virtual model of the repair process and generates the expected parameters of the work order.

[0031] The job deviation analysis unit is used to perform deviation analysis on the real-time job data stream and the expected parameters of the work order, and to identify abnormal job status through data comparison algorithms to generate strategy mismatch early warning signals.

[0032] The mapping relationship optimization unit is used to optimize the resource-capability mapping relationship of the strategy mismatch early warning signal. It adjusts the equipment capability evaluation weight of each repair unit according to the historical operation data recorded in the resource-capability mapping library and generates an updated resource-capability mapping relationship.

[0033] In summary, the multimodal spatiotemporal fusion module in the electrical line fault location and automatic repair system provided in this application achieves panoramic perception of fault information through spatiotemporal alignment and fusion of multimodal data, effectively overcoming the limitations of traditional single data source analysis and laying a solid foundation for subsequent accurate decision-making. The dual-stream collaborative feature extraction module adopts a dual-stream collaborative feature extraction architecture, which can simultaneously mine the electrical and physical characteristics of the fault. This not only enables precise three-dimensional spatial location of the fault point but also forms a deeper characterization of the fault's essence. The repair complexity prediction module, by introducing a multi-factor fusion repair complexity prediction mechanism, can organically combine fault characteristics, environmental constraints, and real-time meteorological conditions to achieve a quantitative assessment of the repair operation's difficulty, providing a scientific basis for resource allocation. The adaptive repair strategy generation module 140 generates adaptive strategies based on a predictive index. It dynamically matches repair resources and plans the optimal operation path according to the actual fault situation, ensuring a precise match between the repair plan and on-site requirements. The repair strategy optimization module establishes a closed-loop optimization mechanism, continuously adjusting system parameters through real-time data feedback. This enables the system to continuously evolve and learn, achieving self-improvement and optimization of the processing strategy. The system provided in this embodiment realizes full-chain technological innovation from perception, cognition, decision-making to optimization, significantly improving the system's adaptability and reliability in multiple scenarios, and providing reliable technical support for achieving precise, efficient, and autonomous fault handling in smart grids.

[0034] To better understand and implement this invention, the following detailed description is provided in conjunction with the accompanying drawings. Attached Figure Description

[0035] Figure 1 A schematic diagram of the structure of a multi-scenario adaptable electrical circuit fault location and automatic repair system provided in an embodiment of this application;

[0036] Figure 2 This is a schematic diagram of the adaptive repair strategy generation module provided in an embodiment of this application. Detailed Implementation

[0037] To facilitate understanding of the present invention, a more complete description will be given below with reference to the accompanying drawings. Preferred embodiments of the invention are shown in the drawings. However, the invention can be implemented in many different forms and is not limited to the embodiments described herein. Rather, these embodiments are provided to provide a thorough and complete understanding of the disclosure of the invention.

[0038] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0039] Please see Figure 1 This embodiment illustrates a multi-scenario adaptable electrical circuit fault location and automatic repair system provided by this application. This embodiment uses the system's application to a terminal as an example for illustration. It is understood that this system can also be applied to a server, and further to a system including both a terminal and a server, and is implemented through interaction between the terminal and the server. Figure 1 As shown, the present invention provides a multi-scenario adapted electrical circuit fault location and automatic repair system 100, which includes a multimodal spatiotemporal fusion module 110, a dual-stream collaborative feature extraction module 120, a repair complexity prediction module 130, an adaptive repair strategy generation module 140, and a fault diagnosis and repair strategy optimization module 150.

[0040] The multimodal spatiotemporal fusion module 110 is used to perform spatiotemporal alignment and fusion processing on electrical and visual data from different timing synchronization sensing devices, unifying fault information from different sources into the same spatiotemporal coordinate system and generating a standardized multimodal fault data cube.

[0041] Specifically, the multimodal spatiotemporal fusion module 110 receives monitoring data from distributed electrical quantity sensors and mobile visual monitoring devices. The electrical quantity sensors are used to collect time-domain and frequency-domain data such as three-phase voltage, three-phase current, active power, reactive power, and harmonic components during line operation. The mobile visual monitoring devices include inspection equipment and fixed monitoring terminals, used to collect images, video frames, and thermal imaging data related to the fault point. All sensing devices integrate time synchronization components to ensure that the raw data carries time stamps and installation location coordinate information, providing a foundation for the spatiotemporal consistency of the data.

[0042] Furthermore, the multimodal spatiotemporal fusion module 110 performs time alignment, employing an interpolation algorithm to unify the time axis of data from different acquisition frequencies. Using the highest acquisition frequency data as a benchmark, lower-frequency data is supplemented. To address time deviations occurring during data transmission, the system corrects them using a pre-set calibration model. This model is constructed based on the statistical laws governing transmission link delays, ensuring that all data remain consistent in the time dimension.

[0043] In the spatial alignment stage, the multimodal spatiotemporal fusion module 110 sets a unified geodetic coordinate system as the spatial reference and maps the initial position coordinates of each sensing device to this reference coordinate system through coordinate transformation operations. For the data collected by the visual monitoring device, the multimodal spatiotemporal fusion module 110 combines the intrinsic and extrinsic parameters of the acquisition device and uses a projection transformation algorithm to convert the pixel coordinates in the image into three-dimensional spatial coordinates, thereby realizing the spatial correlation between visual data and the position information of electrical quantity sensors.

[0044] During data fusion, the multimodal spatiotemporal fusion module 110 employs a confidence-weighted algorithm, assigning corresponding confidence weights based on the measurement accuracy of each sensor. The weight allocation results can be calibrated and adjusted using historical operational data. The system eliminates noise interference in the data through a filtering algorithm, integrates temporal, spatial, and multimodal data information, and generates a standardized multimodal fault data cube. This cube contains three dimensions: time, space, and modality. The time dimension covers a period before and after the fault occurs, the spatial dimension corresponds to three-dimensional coordinate information, and the modality dimension includes two types of data: electrical quantities and visual quantities.

[0045] The dual-stream collaborative feature extraction module 120 is used to perform dual-stream collaborative feature extraction on the multimodal fault data cube. It extracts the electrical characteristics and physical appearance characteristics of the fault through the parallel processing electrical feature extraction branch and visual feature extraction branch, respectively, and generates the accurate three-dimensional coordinates of the fault point and the in-depth fault feature vector.

[0046] Specifically, the dual-stream collaborative feature extraction module 120 constructs a parallel feature extraction architecture, extracting the electrical characteristics and physical manifestations of the fault through two independent processing branches. The electrical feature extraction branch adopts a deep convolutional neural network architecture, with input data consisting of electrical quantity data from a multimodal fault data cube, presented as a combination of time series and electrical quantity types. The network processes the electrical quantity data layer by layer through convolution and pooling operations, extracting fault-related transient components, distortion parameters, impedance changes, and persistent states, forming feature vectors.

[0047] Simultaneously, the dual-stream collaborative feature extraction module 120, combined with the traveling wave localization algorithm, utilizes the propagation characteristics and arrival time differences of the fault traveling wave to calculate the preliminary three-dimensional coordinates of the fault point. The visual feature extraction branch employs an improved deep residual network, with input data consisting of visual data from a multimodal fault data cube, which has already been associated with three-dimensional spatial coordinates. The network extracts the physical appearance features of the fault point through the residual block structure, including shape-related parameters, texture-related parameters, color-related parameters of the defect area, and temperature distribution-related parameters corresponding to thermal imaging, forming another set of feature vectors. The system uses a combination of feature matching and stereo measurement to process multiple frames of visual images and optimize the calculation results of the three-dimensional coordinates of the fault point.

[0048] To improve processing efficiency, the dual-stream collaborative feature extraction module 120 employs a parallel computing architecture to achieve synchronous operation of the two feature extraction branches. After feature extraction is completed, the dual-stream collaborative feature extraction module 120 performs weighted fusion on the preliminary three-dimensional coordinates output by the two branches, assigning appropriate weights based on the characteristics of electrical quantity localization and visual localization to obtain the precise three-dimensional coordinates of the fault point. Simultaneously, the dual-stream collaborative feature extraction module 120 concatenates the feature vectors generated by the two branches, adjusting the vector value range through normalization operations to form a refined fault feature vector. This vector comprehensively integrates the electrical essential characteristics and physical manifestation characteristics of the fault.

[0049] The repair complexity prediction module 130 is used to acquire the geographic environment data of the area where the fault point is located in real time, and calculate the repair complexity prediction index by combining the deepened fault feature vector and the accurate three-dimensional coordinates. The repair complexity prediction index value is generated by weighting and nonlinearly transforming the deepened fault feature vector with environmental accessibility parameters and real-time meteorological parameters.

[0050] Specifically, the repair complexity prediction module 130 calls relevant data from the geographic information system, meteorological monitoring platform, and traffic management platform via interfaces to obtain geographic environmental information, environmental accessibility information, and real-time meteorological information for the area where the fault point is located. Geographic environmental information includes the area's terrain type, road network distribution, road construction status, and the distribution of surrounding obstacles. The system quantifies this information to form terrain correlation coefficients and road network correlation parameters. Environmental accessibility parameters are calculated jointly from the road network distribution and terrain type. Through the calculation of road network correlation parameters and terrain correlation coefficients, a quantitative index characterizing the accessibility of the repair operation is obtained. Real-time meteorological information includes parameters such as wind, precipitation, temperature, and visibility. The repair complexity prediction module 130 quantifies and converts various meteorological parameters to form meteorological influence coefficients.

[0051] Furthermore, the repair complexity prediction module 130 fuses the enhanced fault feature vector with environmental accessibility parameters and real-time meteorological parameters. The fusion process employs an analytic hierarchy process (AHP) to determine the weight distribution of each parameter, with the enhanced fault feature vector holding the primary weight, while environmental accessibility parameters and real-time meteorological parameters hold corresponding weights. The repair complexity prediction module 130 obtains an initial fusion result through weighted summation and then maps this initial fusion result to a repair complexity prediction index using a nonlinear transformation algorithm. This index characterizes the difficulty level of the fault repair operation, with its numerical range corresponding to different repair difficulty intervals, providing a quantitative basis for subsequent repair resource matching and operation planning. Throughout the calculation process, the repair complexity prediction module 130 ensures that the quantification of each parameter conforms to the characteristics of the actual scenario. The design of the fusion and transformation algorithms fully considers the needs of multi-scenario adaptation, enabling the prediction index to accurately reflect the repair difficulty under different fault scenarios.

[0052] The adaptive repair strategy generation module 140 is used to generate adaptive repair strategies based on the predicted index value of repair complexity and the in-depth fault feature vector. It matches repair resources according to the preset resource-capability mapping relationship and plans the operation path to the fault point. It generates a personalized repair work order containing the execution unit, target location and operation procedure and sends it to the repair unit to be dispatched.

[0053] Specifically, the adaptive repair strategy generation module 140 pre-establishes a mapping database between repair resources and operational capabilities. Repair resources encompass three categories: human resources, equipment resources, and material resources. Human resources are categorized based on skill levels, with different skill levels corresponding to different operational capability ranges. Equipment resources include various operational equipment and tools, with different equipment corresponding to different operational scenarios and functions. Material resources include various consumables required for line repair, with different consumables corresponding to different fault repair types. The mapping database is stored in matrix form, with elements representing the degree of adaptability of a certain type of resource to specific fault characteristics. Based on the repair complexity prediction index and the refined fault feature vector, the adaptive repair strategy generation module 140 uses an optimization algorithm to select the resource combination with the highest degree of adaptability from the mapping database.

[0054] During resource matching, the adaptive repair strategy generation module 140 fully considers the repair difficulty corresponding to the predicted index and the fault type represented by the in-depth fault feature vector, ensuring that the selected resource combination can meet the needs of the actual repair operation. In the path planning stage, the adaptive repair strategy generation module 140 uses a path search algorithm based on road network data provided by the geographic information system and real-time traffic information provided by the traffic management platform to plan the operation path to the fault point. During the planning process, various influencing factors are comprehensively considered to ensure that the path enables personnel and equipment to quickly reach the fault site.

[0055] Furthermore, the adaptive repair strategy generation module 140 generates personalized repair work orders, which include execution unit information, target location information, and work procedure information. The execution unit information clearly specifies the personnel, equipment, and materials involved in the repair operation; the target location information clearly specifies the three-dimensional coordinates of the fault point and relevant warning information of the surrounding environment; the work procedure information clearly specifies the operation steps, process requirements, testing procedures, and safety protection requirements for the repair operation. After the work order is generated, the adaptive repair strategy generation module 140 sends the work order to the repair unit terminal to be dispatched via a communication network. The selection of the communication network is based on the network coverage of the work area to ensure that the work order can be transmitted to the execution unit in a timely and reliable manner, providing clear guidance for the implementation of the repair operation.

[0056] The repair strategy optimization module 150 is used to optimize the strategy based on the actual operation data stream fed back by the repair unit. By comparing the difference between the actual operation data and the expected parameters of the work order, the weight parameters of the resource-capability mapping relationship are adjusted by utilizing the online learning capability of the feature extraction network to generate the optimized resource-capability mapping relationship.

[0057] Specifically, the repair strategy optimization module 150 collects actual operation data from the repair unit. This data includes the execution time of the repair operation, consumable consumption, on-site assessment of repair complexity, changes in environmental parameters during the operation, success rate of the repair operation, and deviations during work order execution. The repair strategy optimization module 150 compares and analyzes the collected actual operation data with the expected parameters in the work order, calculating the difference indicators between the two. These difference indicators include parameters related to time deviation, consumable deviation, complexity deviation, and step execution compliance.

[0058] Furthermore, the repair strategy optimization module 150 concatenates these difference indicators with the previously generated in-depth fault feature vector and repair complexity prediction index to form an optimized feature vector. This vector integrates information from multiple aspects, including fault characteristics, prediction results, and actual execution differences. The repair strategy optimization module 150 utilizes the online learning function module in the feature extraction network. This module is built based on an incremental learning algorithm and can dynamically adjust parameters. The system inputs the optimized feature vector into this module, and adjusts the weight parameters in the resource-capacity mapping relationship database through the gradient descent algorithm.

[0059] During the adjustment process, the repair strategy optimization module 150 can employ regularization constraint algorithms to avoid parameter overfitting. Simultaneously, it sets trigger conditions for weight updates, ensuring that weight adjustment is only performed when the difference index reaches a preset standard, thus guaranteeing the stability of the mapping relationship. After weight adjustment, the repair strategy optimization module 150 overwrites the original database with the optimized resource-capability mapping relationship, serving as the resource matching basis for subsequent fault repair operations. Furthermore, the repair strategy optimization module 150 sets global optimization trigger conditions. After completing a certain number of fault repair operations, the module performs global optimization calculations based on all historical operation data within that phase, further adjusting the parameter configuration of the mapping relationship to continuously improve the system's adaptability to different fault scenarios, ensuring that the rationality and effectiveness of the repair strategy are continuously optimized with practical application.

[0060] In summary, the multimodal spatiotemporal fusion module 110 in the electrical line fault location and automatic repair system provided in this application achieves panoramic perception of fault information through spatiotemporal alignment and fusion of multimodal data, effectively overcoming the limitations of traditional single data source analysis and laying a solid foundation for subsequent accurate decision-making. The dual-stream collaborative feature extraction module 120 adopts a dual-stream collaborative feature extraction architecture, which can simultaneously mine the electrical and physical characteristics of the fault. This not only enables precise three-dimensional spatial location of the fault point but also forms a deeper characterization of the fault's essence. The repair complexity prediction module 130, by introducing a multi-factor fusion repair complexity prediction mechanism, can organically combine fault characteristics, environmental constraints, and real-time meteorological conditions to achieve a quantitative assessment of the repair operation's difficulty, providing a scientific basis for resource allocation. The adaptive repair strategy generation module 140 generates adaptive strategies based on a predictive index, dynamically matching repair resources and planning the optimal operation path according to the actual fault situation, ensuring a precise match between the repair plan and on-site requirements. The repair strategy optimization module 150 establishes a closed-loop optimization mechanism, continuously adjusting system parameters through real-time data feedback, enabling the system to have continuous learning capabilities for self-improvement and optimization of the processing strategy. The system provided in this embodiment realizes full-chain technological innovation from perception, cognition, decision-making to optimization, significantly improving the system's adaptability and reliability in multiple scenarios, and providing reliable technical support for achieving precise, efficient, and autonomous fault handling in smart grids.

[0061] In one embodiment, the multimodal spatiotemporal fusion module 110 of the electrical line fault location and automatic repair system 100 provided by the present invention is configured with the following units:

[0062] The electrical signal preprocessing unit is used to preprocess the raw electrical signals collected by the traveling wave sensor group and the fault recorder. It extracts effective fault feature waveforms through filtering and noise cancellation techniques, and generates high-frequency transient current traveling wave sequences and power frequency voltage and current drop waveforms.

[0063] Specifically, the electrical signal preprocessing unit receives raw electrical signals transmitted from a group of traveling wave sensors and a fault recorder. The traveling wave sensors are distributed at key nodes of the electrical line to capture transient traveling wave signals generated when a fault occurs. The fault recorder continuously records voltage and current signals during normal operation and fault conditions. The raw electrical signals collected by both devices contain core information related to the line fault, but are also mixed with interference signals introduced during transmission and noise signals generated by the equipment itself. The electrical signal preprocessing unit filters the raw electrical signals, using targeted filtering techniques to separate the effective components from the interference components. This filtering technique is designed based on the frequency characteristics of the electrical signals, retaining the characteristic frequency components related to the fault and eliminating interference signals of irrelevant frequencies, thus achieving preliminary signal purification.

[0064] Furthermore, the electrical signal preprocessing unit employs noise cancellation technology to further process the filtered signal. This technology constructs a processing model based on the statistical differences between signal and noise, identifying and suppressing remaining noise components by analyzing the amplitude variation and phase characteristics of the signal, thus preventing noise from affecting subsequent feature extraction. After filtering and noise cancellation, the electrical signal preprocessing unit extracts characteristic waveforms from the processed signal, separating waveform components directly related to the fault. Among these, the high-frequency transient current traveling wave sequence corresponds to the transient traveling wave signal generated when the fault occurs, reflecting the initial propagation characteristics of the fault; the power frequency voltage and current drop waveform corresponds to the changes in power frequency electrical quantities under fault conditions, reflecting the impact of the fault on the normal operating parameters of the line. The generation of these two types of characteristic waveforms provides a foundation for subsequent time synchronization processing and feature extraction.

[0065] The image data quality enhancement unit is used to enhance the quality of raw image data collected by inspection drones and fixed monitoring devices. It uses image deblurring and contrast enhancement algorithms to improve visual effects and generate visible light image sequences and infrared thermal imaging sequences.

[0066] Specifically, the image data quality enhancement unit acquires raw image data from inspection drones and fixed monitoring devices. The inspection drones move along a preset route to capture images of different sections of the line. Fixed monitoring devices are installed at key locations along the line to continuously monitor the line status in specific areas. The raw image data acquired by both types of devices may suffer from blurriness or insufficient contrast due to factors such as the shooting environment and equipment performance, affecting the identification of fault-related physical manifestations. The image data quality enhancement unit performs quality enhancement processing on the raw image data, using an image deblurring algorithm to process blurred images. This algorithm is based on the inverse operation principle of image degradation, analyzing the image's blur kernel function to restore detailed information in fault-related areas and eliminate image blurring caused by factors such as camera shake and atmospheric scattering.

[0067] Furthermore, the image data quality enhancement unit uses a contrast enhancement algorithm to adjust the grayscale distribution of the image. By stretching the grayscale range of the image, it strengthens the grayscale difference between the fault area and the background area, making the physical manifestations of the fault clearer and more identifiable. After quality enhancement processing, the image data quality enhancement unit organizes the processed images according to time sequence and spatial location, generating visible light image sequences and infrared thermal imaging sequences. The visible light image sequences visually present the appearance of the line and the physical form of the fault, while the infrared thermal imaging sequences reflect the temperature distribution in various areas of the line, providing high-quality visual data support for the extraction and spatial correlation of the physical manifestations of the fault.

[0068] The spatiotemporal synchronization mapping processing unit is used to perform time synchronization processing on high-frequency transient current traveling wave sequences and power frequency voltage and current drop waveforms, unify the time reference of all electrical signals and extract key waveform features. At the same time, it performs spatial coordinate mapping processing on visible light image sequences and infrared thermal imaging sequences, accurately associates visual data with the line geographic information system, and generates a multimodal fault data cube.

[0069] Specifically, the spatiotemporal synchronization mapping processing unit performs time synchronization processing on the high-frequency transient current traveling wave sequence and the power frequency voltage and current drop waveform. Although the two types of electrical signals are acquired by different devices, they both carry time stamp information. The spatiotemporal synchronization mapping processing unit uses a unified time reference and adjusts the time stamps of the two types of signals through a time calibration algorithm to eliminate time deviations between different acquisition devices and ensure that all electrical signals maintain consistency in the time dimension. After time synchronization is completed, the spatiotemporal synchronization mapping processing unit extracts key waveform features from the synchronized electrical signals. By analyzing parameters such as the propagation speed, amplitude change, and polarity characteristics of the high-frequency transient current traveling wave sequence, and parameters such as the drop amplitude, duration, and recovery trend of the power frequency voltage and current drop waveform, key feature information that can characterize the electrical characteristics of the fault is extracted.

[0070] Simultaneously, the spatiotemporal synchronization mapping processing unit performs spatial coordinate mapping processing on visible light image sequences and infrared thermal imaging sequences. Combining the flight trajectory information of the inspection drone, the installation location information of the fixed monitoring device, and the laying parameters of the line, it uses a spatial coordinate transformation algorithm to convert the pixel coordinates in the images into geographic coordinates, achieving a precise association between visual data and the line's geographic information system, ensuring that each image corresponds to a specific spatial location of the line. The spatiotemporal synchronization mapping processing unit integrates key features of the time-synchronized electrical signals and the spatially mapped visual data to construct a multimodal fault data cube containing time, space, and modal dimensions. The time dimension corresponds to the signal acquisition period before and after the fault occurred, the spatial dimension corresponds to the line's geographic coordinate information, and the modal dimension includes electrical and visual quantity features. This data cube achieves structured integration of fault information from different sources.

[0071] In one embodiment, the dual-stream collaborative feature extraction module 120 of the electrical line fault location and automatic repair system 100 provided by the present invention is configured with the following units:

[0072] The electrical multi-scale feature extraction unit is used to perform multi-scale feature extraction processing on electrical quantity data in the multimodal fault data cube. It mines fault features in different frequency bands through wavelet packet transform and deep convolutional network to generate electrical feature vectors containing transient and steady-state characteristics.

[0073] Specifically, the electrical multi-scale feature extraction unit calls upon the electrical quantity data in the multi-modal fault data cube. This data includes high-frequency transient current traveling wave sequences and power frequency voltage and current drop waveforms after time synchronization processing, integrating electrical signal change information before and after the fault occurrence, covering the relevant features of transient and steady-state processes. The electrical multi-scale feature extraction unit performs wavelet packet transform processing on the electrical quantity data. This transform technique can decompose the electrical signal into multi-scale components, breaking down the original electrical signal into sub-signals of different frequency bands. Each frequency band sub-signal corresponds to the characteristic performance of different stages of the fault, with the high-frequency sub-signals mainly reflecting the characteristics of the transient fault process and the low-frequency sub-signals mainly reflecting the characteristics of the steady-state fault process.

[0074] Through wavelet packet transform, the electrical multi-scale feature extraction unit achieves comprehensive coverage of the full-frequency band features of electrical signals, avoiding feature omissions caused by single-scale analysis. The unit inputs the decomposed frequency band sub-signals into a deep convolutional network. The network extracts features from each frequency band sub-signal through layer-by-layer convolution operations. The first convolutional layer captures the local basic features of the signal, while subsequent convolutional layers gradually mine deeper abstract features through superposition operations. Simultaneously, pooling operations are used to filter and compress the extracted features, retaining key feature information and reducing data dimensionality. The unit integrates the frequency band features output by the deep convolutional network, associating and integrating the transient features corresponding to the high-frequency band with the steady-state features corresponding to the low-frequency band, forming an electrical feature vector containing both transient and steady-state characteristics. This vector comprehensively covers the multi-dimensional feature information of faults at the electrical level.

[0075] The visual attention enhancement unit is used to enhance the visual data in the multimodal fault data cube through an attention mechanism. It uses the spatial attention module to focus on the visual anomalies in the fault area and generate a visual feature vector with spatial positioning capabilities.

[0076] Specifically, the visual attention enhancement unit extracts visual data from the multimodal fault data cube. This data includes visible light image sequences and infrared thermal imaging sequences processed by spatial coordinate mapping, which can intuitively present the physical morphology and abnormal temperature distribution of the fault area. The visual attention enhancement unit performs attention mechanism enhancement processing on the visual data, with the core employing a spatial attention module to enhance the features of the image data. The spatial attention module analyzes the correlation between image pixels and calculates the weight coefficient of each pixel related to fault identification. The allocation of weight coefficients is based on information such as pixel grayscale changes, texture distribution, and temperature differences. Pixels in fault-related areas receive higher weight coefficients, while pixels in background and irrelevant areas receive lower weight coefficients.

[0077] By allocating weighting coefficients, the visual attention enhancement unit focuses on the visual anomalies of the fault area, strengthens the feature differences between the fault area and the background area, and suppresses the interference of irrelevant background information on fault feature extraction. During the attention mechanism enhancement process, the visual attention enhancement unit processes visible light image sequences and infrared thermal imaging sequences separately. The processing of the visible light image sequences focuses on the morphological and structural anomalies of the fault area, while the processing of the infrared thermal imaging sequences focuses on the temperature distribution anomalies of the fault area. The processing results of the two types of sequences complement each other, comprehensively capturing the visual appearance features of the fault. After processing, the visual attention enhancement unit structurally integrates the enhanced visual features, combining them with the spatial coordinate information of the image to generate a visual feature vector with spatial positioning capabilities. This vector not only contains the visual anomalies of the fault but also associates with the spatial location information of the fault, providing visual support for spatial positioning after cross-modal fusion.

[0078] The cross-modal feature fusion unit is used to perform cross-modal feature fusion processing on electrical feature vectors and visual feature vectors. Through feature splicing and fully connected layers, it learns the correlation between modes and generates a comprehensive fault feature vector and accurate three-dimensional coordinates that reflect the fault characteristics.

[0079] Specifically, the cross-modal feature fusion unit acquires electrical feature vectors and visual feature vectors. These two types of vectors represent fault characteristics from the perspectives of electrical essence and visual appearance, respectively, and are inherently related. The cross-modal feature fusion unit performs feature concatenation processing on the two types of feature vectors, combining the electrical feature vectors and visual feature vectors in dimensional order to form an initial cross-modal feature set. This set contains both electrical and visual fault characteristic information, but the two types of information are in an independent state and do not form an effective correlation.

[0080] Furthermore, the cross-modal feature fusion unit inputs the initial cross-modal feature set into the fully connected layer. The fully connected layer learns the correlations between modalities through multi-layer neural network operations, uncovering the inherent mapping patterns between electrical and visual features, such as the correspondence between abnormal fault electrical parameters and visual morphological anomalies, and the correlation patterns between transient electrical features and abnormal temperature distribution. During the learning process, the fully connected layer performs nonlinear transformations and dimensional adjustments on the concatenated features, eliminating heterogeneity between different modal features and achieving deep feature fusion.

[0081] After the fusion process is completed, the cross-modal feature fusion unit outputs two results: First, a deepened fault feature vector that comprehensively reflects the fault characteristics. This vector integrates the electrical essential characteristics, visual appearance characteristics, and intermodal correlation characteristics of the fault, and can comprehensively and accurately characterize the fault state. Second, precise three-dimensional coordinates. The generation of these coordinates is based on the spatial coordinate information in the fused cross-modal features and multimodal fault data cube. Through precise mapping between features and spatial positions, combined with the traveling wave propagation characteristics in the electrical features and the spatial positioning information in the visual features, the precise calculation of the three-dimensional coordinates of the fault point is achieved.

[0082] In one embodiment, the repair complexity prediction module 130 of the electrical circuit fault location and automatic repair system 100 provided by the present invention is configured with the following units:

[0083] The environmental accessibility assessment unit is used to acquire geographic environmental data of the area where the fault point is located in real time. It processes the geographic environmental data of the area where the fault point is located in real time, and obtains terrain slope, vegetation density and traffic accessibility parameters by querying the geographic information system to generate an environmental accessibility assessment vector.

[0084] Specifically, the environmental accessibility assessment form uses the precise three-dimensional coordinates of the fault point as an index to establish a communication connection with the Geographic Information System (GIS) and initiates a data query request through a preset data interaction protocol. Based on the received coordinate information, the GIS locates the specific area where the fault point is located and extracts relevant data on terrain slope, vegetation density, and traffic accessibility for that area. Terrain slope data reflects the inclination of the regional surface; the environmental accessibility assessment form processes this data using digital elevation model data stored in the GIS, transforming the undulations of the regional terrain into quantifiable parameters. Vegetation density data is generated based on land cover classification data in the GIS, by statistically analyzing the distribution of vegetation cover within a certain range around the fault point. Traffic accessibility parameters are calculated comprehensively by combining road network distribution data and traffic facility layout data within the area, reflecting the ease with which repair personnel and equipment can reach the fault point.

[0085] Furthermore, the environmental accessibility assessment system standardizes the acquired data on terrain slope, vegetation density, and traffic accessibility to eliminate dimensional differences between different data types and ensure that all types of data are represented on a unified dimension. The system then integrates the standardized data into a structured manner according to a preset order to construct an environmental accessibility assessment vector that includes terrain features, vegetation distribution features, and traffic features. This vector comprehensively reflects the impact of the geographical environment of the fault location area on the remediation work.

[0086] The meteorological impact assessment unit is used to collect and process meteorological monitoring data in the area where the fault point is located in real time. It obtains real-time parameters such as wind speed, precipitation and temperature by connecting to the meteorological data interface and generates a meteorological impact assessment vector.

[0087] Specifically, the meteorological impact assessment unit establishes a stable data transmission link with the meteorological monitoring platform through a pre-defined meteorological data interface, and acquires real-time meteorological monitoring data for the area based on the geographical coordinates of the fault point. During data acquisition, the meteorological impact assessment unit initiates data requests at fixed time intervals to ensure the real-time and continuous nature of the meteorological data. The collected meteorological data includes wind speed, precipitation, and temperature-related parameters. Wind speed data reflects the intensity of air movement within the area, precipitation data reflects the occurrence of precipitation within the area, and temperature data reflects the ambient temperature within the area.

[0088] Furthermore, the meteorological impact assessment unit verifies the validity of all collected meteorological data, eliminating invalid data caused by transmission interference or equipment malfunctions to ensure the reliability of the input data. After verification, the unit quantifies and transforms the various meteorological data, converting meteorological information in different forms into quantified parameters of a unified dimension. Wind speed data is transformed based on the intensity characteristics of airflow, precipitation data based on the intensity and duration of precipitation, and temperature data based on the range of ambient temperature. The unit then combines the quantified wind speed, precipitation, and temperature parameters according to predefined rules to form a meteorological impact assessment vector. This vector centrally reflects the constraints of real-time meteorological conditions on the remediation operations.

[0089] The repair complexity prediction unit is used to perform multi-factor fusion calculations on the in-depth fault feature vector, environmental accessibility assessment vector, and meteorological impact assessment vector. It evaluates the difficulty of repair operations through weighted summation and nonlinear activation function, and generates a repair complexity prediction index value.

[0090] Specifically, the repair complexity prediction unit calls upon the enhanced fault feature vector, environmental accessibility assessment vector, and meteorological impact assessment vector. These three types of vectors provide quantitative data from three dimensions: fault characteristics, geographical environmental conditions, and meteorological environmental conditions, respectively, collectively constituting the core factors for repair complexity assessment. The repair complexity prediction unit performs dimension unification processing on the three types of vectors, using a feature dimension adjustment algorithm to ensure consistency in their dimensions, laying the foundation for subsequent fusion calculations. Subsequently, the system performs a weighted summation operation, assigning corresponding weights to each vector based on the importance of the influencing factors corresponding to the three types of vectors in the repair complexity assessment. The weight allocation is determined based on engineering practice rules in fault handling, ensuring that the weight settings conform to actual application scenarios.

[0091] Furthermore, the repair complexity prediction unit multiplies the corresponding feature elements in each vector with their assigned weights, and then sums all the product results to obtain the initial fusion result. After the initial fusion result is generated, the repair complexity prediction unit inputs it into a preset nonlinear activation function. This function maps the initial fusion result through nonlinear transformation, strengthening the correlation between different influencing factors and accurately depicting the changing pattern of repair operation difficulty under the combined effect of various factors. After nonlinear activation processing, the repair complexity prediction unit outputs the final repair complexity prediction index value. This index value is a comprehensive quantitative representation of the repair operation difficulty, fully reflecting the combined impact of fault characteristics, geographical environment, and meteorological conditions on the repair operation.

[0092] In one embodiment, such as Figure 2 As shown, the adaptive repair strategy generation module 140 of the electrical circuit fault location and automatic repair system 100 provided by the present invention is configured with the following units:

[0093] The repair resource matching unit is used to intelligently match repair resources based on the repair complexity prediction index value. It selects suitable repair equipment combinations from the resource library according to a preset threshold range and generates the optimal repair unit set.

[0094] Specifically, the repair resource matching unit calls the repair complexity prediction index value to perform intelligent matching of repair resources. A repair resource library is pre-established, categorizing and archiving various repair equipment according to operational functions and application scenarios. The stored information includes equipment technical parameters, operational scope, and collaborative operation compatibility information, forming a structured resource storage system. The repair resource matching unit incorporates threshold interval division rules based on common practical rules of repair operations, dividing the range of changes in the repair complexity prediction index value into different intervals. Each interval corresponds to a set of suitable repair equipment types. The system compares the acquired repair complexity prediction index value with the preset threshold interval to determine the target interval and initiates an equipment screening request in the resource library using the target interval as the search condition. During the screening process, the repair resource matching unit performs quantitative evaluation using a resource adaptability calculation formula, which is:

[0095]

[0096] in, Indicates the overall compatibility of the equipment combination. This represents the weight coefficient for the i-th type of device. This represents the compatibility parameter between the i-th type of equipment and the fault scenario. The weighting coefficient is determined based on the functional importance of the equipment in the repair operation, and the compatibility parameter is calculated through the correspondence between the equipment's technical parameters and the fault requirements. The repair resource matching unit combines equipment collaborative operation compatibility information to exclude combinations with functional conflicts or those that cannot operate together, and refers to the current status information of the equipment in the resource library to ensure that the selected equipment is in an available state. The repair resource matching unit sorts the equipment combinations that meet the conditions according to their comprehensive suitability, and finally selects the equipment combination with the highest comprehensive suitability to generate the optimal repair unit set.

[0097] The three-dimensional path planning unit is used to perform three-dimensional path planning based on the pre-set resource-capacity mapping relationship and the deepened fault feature vector, considering the fault location and environmental accessibility parameters. It comprehensively considers terrain obstacle avoidance and equipment operation capability limitations to calculate the optimal travel trajectory and generate a safe and efficient operation path scheme.

[0098] Specifically, the 3D path planning unit retrieves pre-set resource-capability mapping relationships and refined fault feature vectors, associates them with acquired fault point location information and environmental accessibility parameters, and performs 3D path planning. The resource-capability mapping relationship includes information such as the operational capability boundaries, travel restrictions, and applicable terrain ranges of various repair equipment. The 3D path planning unit uses this mapping relationship to clarify the operational capability constraints corresponding to the currently matched set of repair units. Information from the refined fault feature vectors assists the system in judging the distribution status of routes around the fault point and potential operational risks. Using the precise 3D coordinates of the fault point as the target point, the 3D path planning unit, combined with terrain data and obstacle distribution data provided by the geographic information system, constructs a 3D terrain model of the area where the fault point is located. During the path planning process, the 3D path planning unit can use a path comprehensive cost calculation formula to evaluate different path options. The formula is:

[0099]

[0100] in, Indicates the total cost of the path. This represents the path length parameter. This indicates the difficulty parameter for obstacle avoidance. Indicates device compatibility parameters. , , These are the weighting coefficients for each parameter. The path length parameter reflects the spatial span of the path, the obstacle avoidance difficulty parameter is calculated based on the obstacle types and distribution range identified by the 3D terrain model, and the equipment adaptability parameter is determined based on the travel capability and operating radius requirements of the repair equipment.

[0101] Furthermore, the three-dimensional path planning unit uses a path search algorithm to bypass impassable obstacle areas, strictly adheres to equipment operation capacity limitations, evaluates the comprehensive cost of different path schemes through multiple rounds of path iteration calculations, eliminates path schemes with excessive comprehensive costs, determines the optimal travel trajectory, generates a safe and efficient operation path scheme, and clarifies the travel route, nodes, avoidance areas, and key operation prompts of the repair unit from the starting position to the fault point.

[0102] The repair work order synthesis unit is used to automatically synthesize work orders from repair unit sets and work path plans. It combines fault type and repair complexity to generate detailed operation instructions and emergency plans, and generates executable personalized repair work orders.

[0103] Specifically, the repair work order synthesis unit integrates the repair unit set and the operation path plan, calls the fault type identification results and the repair complexity prediction index value, and automatically synthesizes the work order. It performs structured processing on various input data, and classifies and integrates the equipment details in the repair unit set, the route information in the operation path plan, the repair process requirements corresponding to the fault type, and the operation process specifications corresponding to the repair complexity according to the preset work order structure framework.

[0104] During the operation instruction generation phase, the repair work order synthesis unit formulates detailed step-by-step operation instructions based on the fault type and equipment characteristics of the repair unit, combined with the standard repair operation process. In the emergency plan development process, the repair work order synthesis unit refers to the risk level corresponding to the repair complexity prediction index value, and, combined with the geographical environment and meteorological conditions of the fault location area, identifies potential emergencies during the operation and formulates corresponding countermeasures. Preferably, the repair work order synthesis unit ensures comprehensive work order information through a work order integrity index formula, the formula being:

[0105]

[0106] in, This represents the work order completeness index. This indicates the number of operation instructions already included. This indicates the total number of operation instructions required. This indicates the detail parameter of the operation command. This indicates the number of emergency response plans that have been developed. This indicates the total number of emergency response plans required. This indicates the feasibility parameter of the emergency plan. The detail parameter of the operation instructions is calculated based on the completeness of the instruction steps, while the feasibility parameter of the emergency plan is determined based on the degree of matching between the response measures and the emergency situation. The repair work order synthesis unit organically integrates the structured information, detailed operation instructions, and emergency plan, formats them according to the standardized work order format, and generates an executable, personalized repair work order.

[0107] In one embodiment, the repair strategy optimization module 150 of the electrical circuit fault location and automatic repair system 100 provided by the present invention is configured with the following units:

[0108] The operation status monitoring unit is used to monitor and process the operation status data fed back by the repair unit in real time. It collects information such as the robot arm posture, operation progress and environmental changes through sensors to generate a real-time operation data stream.

[0109] Specifically, the operation status monitoring unit establishes a continuous data transmission link with various sensors. These sensors are distributed at key parts of the repair equipment and monitoring points in the work environment to collect core data during the operation. The collected information includes data related to the robotic arm's posture, operation progress, and environmental changes. The robotic arm posture data reflects the spatial position and motion status of the operating mechanism; the operation progress data records the completion status of each stage of the repair operation; and the environmental change data captures the dynamic changes in the geographical environment and weather conditions during the operation. The operation status monitoring unit preprocesses the raw data collected by the sensors, eliminating invalid data using a data validity verification formula:

[0110]

[0111] in, This indicates valid data after verification. This represents the error correction factor. This represents the raw data collected by the sensor. This represents the data reference baseline value. The error correction coefficient is determined based on the sensor's measurement characteristics, and the data reference baseline value is based on the standard state parameters set for the repair operation. The operation status monitoring unit integrates the verified valid data in a structured manner according to time sequence and data type, arranges data entries according to the preset data format specifications, and forms a real-time operation data stream containing timestamps, data type identifiers, and specific parameter values. This data stream completely records the dynamic execution process of the repair operation.

[0112] The digital twin simulation unit is used to perform digital twin simulation based on the deepened fault feature vector and environmental accessibility assessment vector. It predicts the operation parameters of each stage by establishing a virtual model of the repair process and generates the expected parameters of the work order.

[0113] Specifically, the digital twin simulation unit retrieves the enhanced fault feature vector and environmental accessibility assessment vector for digital twin simulation. Based on the precise three-dimensional coordinates of the fault point and relevant information in the multimodal fault data cube, a virtual model of the repair process is constructed. This model replicates the line structure, geographical environment, and physical characteristics of the repair equipment at the fault point, achieving a digital mapping of the repair operation scenario. Further, the digital twin simulation unit inputs the enhanced fault feature vector and environmental accessibility assessment vector into the virtual model, and uses simulation algorithms to deduce the execution process of each stage of the repair operation, predicting the key parameters required for the operation. The prediction process uses a parameter mapping formula for quantitative calculation, the formula being:

[0114]

[0115] in, Indicates the expected parameters of the work order. Represents the fault feature mapping matrix. This represents a deeper fault feature vector. Represents the environmental factor mapping matrix. This represents the environmental accessibility assessment vector. The fault feature mapping matrix characterizes the correlation between fault features and operational parameters, while the environmental factor mapping matrix characterizes the influence of environmental conditions on operational parameters. The system optimizes prediction results through multiple rounds of iterative simulation using a virtual model, generating expected work order parameters covering operation duration, equipment operating parameters, consumable consumption, and quality acceptance standards.

[0116] The job deviation analysis unit is used to perform deviation analysis on the real-time job data stream and the expected parameters of the work order, identify abnormal job status through data comparison algorithms, and generate strategy mismatch early warning signals.

[0117] Specifically, the job deviation analysis unit acquires real-time job data streams and expected work order parameters, performs dimensional unification and format standardization on both types of data to ensure data comparability, and correlates each parameter in the real-time job data stream with its corresponding item in the expected work order parameters one by one. The job deviation analysis unit uses a data comparison algorithm to calculate the degree of deviation for each parameter, and accurately represents it through a deviation quantification formula, which is:

[0118]

[0119] in, This represents the relative deviation value. Represents a real-time job data vector. This represents the expected parameter vector for the work order. This represents L2 norm calculation. The relative deviation value reflects the degree of deviation between real-time operational data and expected parameters. The operational deviation analysis unit compares the calculated relative deviation value with a preset deviation threshold to identify abnormal operational states that exceed the threshold range. Abnormal operational states include operational delays, deviations in equipment operating parameters, and environmental impacts exceeding expectations. The operational deviation analysis unit classifies and identifies various abnormal states, integrates abnormal information to generate a strategy mismatch early warning signal, which includes key information such as the type of abnormality, the time of occurrence, the degree of deviation, and related operational processes.

[0120] The mapping relationship optimization unit is used to optimize the resource-capability mapping relationship of the strategy mismatch early warning signal. It adjusts the equipment capability evaluation weight of each repair unit according to the historical operation data recorded in the resource-capability mapping library and generates an updated resource-capability mapping relationship.

[0121] Specifically, the mapping relationship optimization unit receives a strategy mismatch warning signal and retrieves historical operation data stored in the resource-capability mapping library to optimize the mapping relationship. The resource-capability mapping library records the fault characteristics, environmental conditions, resource allocation schemes, and operation effect data of each repair operation, forming a structured historical database. The mapping relationship optimization unit uses the abnormal information in the strategy mismatch warning signal as an index to retrieve resource allocation cases under similar scenarios in the historical operation data, and analyzes the compatibility between the equipment capabilities of the repair units and the operation requirements in these cases. The mapping relationship optimization unit dynamically adjusts the equipment capability evaluation weights of each repair unit using a weight adjustment formula, which is:

[0122]

[0123] in, This indicates the updated equipment capability assessment weights. This indicates the weighting of the original equipment capability assessment. This represents the learning rate parameter. Indicates the deviation value. This represents the historical fit parameter. The learning rate parameter controls the step size of the weight adjustment, and the historical fit parameter is calculated based on the operational results of similar cases. After the mapping relationship optimization unit adjusts the equipment capability evaluation weights of all repair units, it reconstructs the resource-capability mapping relationship to ensure that the mapping relationship can adapt to the deviation in actual operation, generates the updated resource-capability mapping relationship, and stores it in the resource-capability mapping library.

[0124] In the description of this specification, references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of those different embodiments or examples.

[0125] For the device embodiments, since they basically correspond to the method embodiments, the relevant parts can be referred to in the description of the method embodiments. The device embodiments described above are merely illustrative. The components described as separate parts may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this disclosure according to actual needs. Those skilled in the art can understand and implement this without creative effort.

[0126] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any person skilled in the art can easily conceive of various variations or substitutions within the technical scope disclosed in this application, and these should all be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A multi-scenario adaptable electrical circuit fault location and automatic repair system, characterized in that, The system is configured with the following modules: The multimodal spatiotemporal fusion module is used to perform spatiotemporal alignment and fusion processing on electrical and visual data from different timing synchronization sensing devices, unifying fault information from different sources into the same spatiotemporal coordinate system and generating a standardized multimodal fault data cube. The dual-stream collaborative feature extraction module is used to perform dual-stream collaborative feature extraction on the multimodal fault data cube. It extracts the electrical characteristics and physical appearance characteristics of the fault through the parallel processing electrical feature extraction branch and visual feature extraction branch, respectively, and generates the accurate three-dimensional coordinates of the fault point and the in-depth fault feature vector. The repair complexity prediction module is used to acquire the geographic environment data of the area where the fault point is located in real time, and calculate the repair complexity prediction index by combining the deepened fault feature vector and the precise three-dimensional coordinates. The repair complexity prediction index value is generated by weighted fusion and nonlinear transformation of the deepened fault feature vector with environmental accessibility parameters and real-time meteorological parameters. An adaptive repair strategy generation module is used to generate an adaptive repair strategy based on the repair complexity prediction index value and the deepened fault feature vector. It matches repair resources according to the preset resource-capability mapping relationship and plans the operation path to the fault point. It generates a personalized repair work order containing the execution unit, target location and operation program and sends it to the repair unit to be dispatched. The repair strategy optimization module is used to optimize the strategy based on the actual operation data stream fed back by the repair unit. By comparing the difference between the actual operation data and the expected parameters of the work order, it uses the online learning capability of the feature extraction network to adjust the weight parameters of the resource-capability mapping relationship and generate the optimized resource-capability mapping relationship.

2. The system according to claim 1, characterized in that, The multimodal spatiotemporal fusion module is configured with the following units: The electrical signal preprocessing unit is used to preprocess the raw electrical signals collected by the traveling wave sensor group and the fault recorder, extract effective fault feature waveforms through filtering and noise cancellation techniques, and generate high-frequency transient current traveling wave sequences and power frequency voltage and current drop waveforms. The image data quality enhancement unit is used to enhance the quality of the raw image data collected by inspection drones and fixed monitoring devices. It uses image deblurring and contrast enhancement algorithms to improve the visual effect and generate visible light image sequences and infrared thermal imaging sequences. The spatiotemporal synchronization mapping processing unit is used to perform time synchronization processing on the high-frequency transient current traveling wave sequence and the power frequency voltage and current drop waveform, unify the time reference of all electrical signals and extract key waveform features. At the same time, it performs spatial coordinate mapping processing on the visible light image sequence and the infrared thermal imaging sequence, accurately associates the visual data with the line geographic information system, and generates a multimodal fault data cube.

3. The system according to claim 1, characterized in that, The dual-stream collaborative feature extraction module is configured with the following units: The electrical multi-scale feature extraction unit is used to perform multi-scale feature extraction processing on the electrical quantity data in the multimodal fault data cube. It mines fault features in different frequency bands through wavelet packet transform and deep convolutional network to generate electrical feature vectors containing transient and steady-state characteristics. The visual attention enhancement unit is used to perform attention mechanism enhancement processing on the visual data in the multimodal fault data cube, and uses the spatial attention module to focus on the visual anomaly features of the fault area to generate a visual feature vector with spatial positioning capability. The cross-modal feature fusion unit is used to perform cross-modal feature fusion processing on the electrical feature vector and the visual feature vector. Through feature splicing and learning the correlation between modes through a fully connected layer, it generates a comprehensive fault feature vector and accurate three-dimensional coordinates that reflect the fault characteristics.

4. The system according to claim 1, characterized in that, The repair complexity prediction module is configured with the following units: The environmental accessibility assessment unit is used to acquire geographic environmental data of the area where the fault point is located in real time, process the geographic environmental data of the area where the fault point is located in real time, obtain terrain slope, vegetation density and traffic accessibility parameters by querying the geographic information system, and generate an environmental accessibility assessment vector. The meteorological impact assessment unit is used to collect and process meteorological monitoring data in the area where the fault point is located in real time. It obtains real-time parameters such as wind speed, precipitation and temperature by connecting to the meteorological data interface and generates a meteorological impact assessment vector. The repair complexity prediction unit is used to perform multi-factor fusion calculation on the deepened fault feature vector, the environmental accessibility assessment vector, and the meteorological impact assessment vector, and evaluate the difficulty of the repair operation through weighted summation and nonlinear activation function to generate a repair complexity prediction index value.

5. The system according to claim 1, characterized in that, The adaptive repair strategy generation module is configured with the following units: The repair resource matching unit is used to perform intelligent repair resource matching processing on the repair complexity prediction index value, and select suitable repair equipment combinations from the resource library according to the preset threshold range to generate the optimal repair unit set. The three-dimensional path planning unit is used to perform three-dimensional path planning based on the preset resource-capacity mapping relationship and the deepened fault feature vector, considering the fault location and environmental accessibility parameters, and comprehensively considering terrain obstacle avoidance and equipment operation capacity limitations to calculate the optimal travel trajectory and generate a safe and efficient operation path scheme. The repair work order synthesis unit is used to automatically synthesize work orders from the set of repair units and the work path scheme, and generate detailed operation instructions and emergency plans by combining the fault type and repair complexity, thus generating executable personalized repair work orders.

6. The system according to any one of claims 1-5, characterized in that, The repair strategy optimization module is configured with the following units: The operation status monitoring unit is used to monitor and process the operation status data fed back by the repair unit in real time. It collects information such as the robot arm posture, operation progress and environmental changes through sensors to generate real-time operation data stream. The digital twin simulation unit is used to perform digital twin simulation based on the deepened fault feature vector and the environmental accessibility assessment vector, and to predict the operation parameters of each stage by establishing a virtual model of the repair process, and generate expected parameters for the work order. The job deviation analysis unit is used to perform deviation analysis processing on the real-time job data stream and the expected parameters of the work order, identify abnormal job status through data comparison algorithm, and generate a strategy mismatch early warning signal. The mapping relationship optimization unit is used to optimize the resource-capability mapping relationship of the strategy mismatch warning signal, adjust the equipment capability evaluation weight of each repair unit according to the historical operation data recorded in the resource-capability mapping library, and generate an updated resource-capability mapping relationship.

Images