Ground loop resistance detection and fault location method and system integrated with edge computing

Abstract

The present invention discloses a ground loop resistance detection and fault location method and system integrated with edge computing, which relates to the field of ground detection technology. The method includes: performing key node analysis of the ground loop, deploying edge computing units, configuring mapped monitoring sensors, including high-frequency transient traveling wave detectors, broadband impedance spectrum measurement sensors, and synchronous clock units; during the normal operation of the power grid device, using the monitoring sensors to monitor and perform time synchronization alignment of the monitoring data sets, and reconstructing the continuous situation field of the ground loop through interpolation of the edge computing unit, and the continuous situation field is constructed based on three-dimensional fault characteristics; using the monitoring sensors to monitor the power grid device in real time, and based on the real-time monitoring results and the continuous situation field, combined with the local height gradient characteristics and extreme value area characteristics, performing fault source inversion and establishing a fault location result. This achieves the technical effects of reducing response delay, optimizing load distribution, and improving fault location accuracy.

Description

Technical Field

[0001] The present invention relates to the field of grounding detection technology, and in particular to a grounding loop resistance detection and fault location method and system integrated with edge computing. Background Art

[0002] The monitoring and fault location of ground loops are key links in ensuring the stable operation of power grids. With the continuous expansion of the scale of distribution networks and the large-scale access of distributed power sources, existing technologies mainly rely on centralized data centers to uniformly process the monitoring data of power grid devices, transmit field monitoring data to central servers through overhead lines or optical fiber communications, and then use mathematical models to perform fault analysis. However, there are the following major defects: First, the separation of monitoring data acquisition and processing leads to poor time synchronization. Especially in high-frequency transient traveling wave detection scenarios, the time deviation between the data acquisition end and the processing end may exceed the millisecond level, seriously affecting the accuracy of fault feature extraction. Second, traditional fault location methods are mostly based on two-dimensional impedance spectrum characteristics and cannot fully capture the three-dimensional spatial fault characteristics of ground loops, resulting in insufficient fault source inversion accuracy under complex power grid topology structures. Third, when facing large-scale distributed monitoring points, the centralized processing architecture has limited data transmission bandwidth and the risk of single-point failure, making it difficult to meet the real-time and reliability requirements of smart grids. Summary of the Invention

[0003] The present invention provides a ground loop resistance detection and fault location method and system integrated with edge computing to solve the technical problems in the prior art such as large data synchronization deviation, heavy computing and transmission load, and insufficient fault location accuracy, thereby achieving the technical effects of reducing response delay, optimizing load distribution, and improving fault location accuracy.

[0004] In a first aspect, the present invention provides a ground loop resistance detection and fault location method integrating edge computing, wherein the ground loop resistance detection and fault location method integrating edge computing includes:

[0005] Perform key node analysis of the ground loop, deploy edge computing units based on the key node analysis results, and configure monitoring sensors mapped to the edge computing units. The monitoring sensors include a high-frequency transient traveling wave detector, a broadband impedance spectrum measurement sensor, and a synchronous clock unit.

[0006] When the power grid device is operating normally, the monitoring sensor is used to perform power grid device monitoring. After the monitoring data set is time-synchronized and aligned based on the synchronous clock unit, the continuous situation field of the ground loop is reconstructed through interpolation by the edge computing unit. The continuous situation field is constructed based on the three-dimensional fault characteristics.

[0007] The monitoring sensor is used to monitor the power grid device in real time, and the fault source is inverted based on the local height gradient characteristics and extreme value area characteristics through the real-time monitoring results and the continuous situation field to establish fault location.

[0008] In a feasible implementation, reconstructing the continuous situation field of the ground loop by interpolation of the edge computing unit includes:

[0009] After using the synchronous clock unit to synchronize the monitoring data set, the time-aligned monitoring data set is sent to the corresponding edge computing unit. After the edge computing units form a computing local area network:

[0010] The spatial propagation feature extraction of high-frequency transient traveling waves is performed to establish spatial propagation features, which include inter-node propagation delay features, attenuation coefficient features, and local reflection features.

[0011] Spectrum response feature extraction of the broadband impedance spectrum is performed to establish spectrum response features, wherein the spectrum response features include amplitude-frequency variation features, phase-frequency variation features, and frequency segmentation feature vectors.

[0012] Temporal evolution features are extracted from monitoring data sets of multiple key nodes to establish temporal evolution features, which include synchronous evolution features, node association features, and mutation features.

[0013] A three-dimensional fault feature is constructed using the spatial propagation feature, the spectrum response feature, and the time domain evolution feature. After interpolating the three-dimensional fault feature, a continuous state field of the ground loop is reconstructed.

[0014] In a feasible implementation, reconstructing the continuous situation field of the ground loop after interpolating the three-dimensional fault features includes:

[0015] Establish connection edges at all key nodes based on the dependency relationships of the key nodes.

[0016] The edge weights are configured based on the spatial distance and feature difference of the edges.

[0017] After establishing the strength constraint and direction constraint of feature propagation transition based on the connection edge weight, the three-dimensional fault features are traversed to perform feature interpolation to reconstruct the continuous situation field of the ground loop.

[0018] In a feasible implementation, performing key node analysis of a ground loop and deploying an edge computing unit based on the key node analysis results includes:

[0019] Analyze all routable nodes to construct a ground loop.

[0020] After configuring the fusion multidimensional indicators, the node importance evaluation of all deployable nodes is performed using the fusion multidimensional indicators, which include electrical contribution index, propagation sensitivity index, spectrum heterogeneity index, time domain evolution coupling index, and fault evolution potential index.

[0021] The node importance evaluation results are used for proportional screening. After node distribution compensation is performed based on the proportional screening results, key nodes are established and edge computing units are deployed at the key nodes.

[0022] In a feasible implementation, the fault source inversion based on local height gradient characteristics and extreme area characteristics is performed through the real-time monitoring results and the continuous situation field to establish the fault location, including:

[0023] After the real-time monitoring results are time-synchronized and aligned, the time-synchronized real-time monitoring results are superimposed on the continuous situation field according to time slices to form a dynamic situation increment map.

[0024] The local gradient is calculated on the dynamic situation increment map, and the gradient modulus is established.

[0025] The gradient modulus is used to screen high gradient areas and establish local high gradient features.

[0026] Local neighborhood search is used to identify local maximum and local minimum points based on local gradients to establish extreme area features.

[0027] The fault source is inverted using the local height gradient characteristics and extreme value area characteristics.

[0028] In a feasible implementation, after establishing the gradient modulus, the following steps are included:

[0029] Gets the fixed threshold for matching the current monitoring scene.

[0030] The mean and standard deviation of the global gradient field are calculated according to the gradient modulus, and the sensitivity adjustment factor is constructed using the calculation results of the mean and standard deviation.

[0031] After the sensitivity adjustment factor is used to perform focus enhancement on the matching fixed decision threshold, high gradient regions are screened using the focus enhanced matching fixed decision threshold.

[0032] In a feasible implementation, the performing of fault source inversion based on the local height gradient feature and the extreme value area feature includes:

[0033] After locating the fault source, the synchronous clock unit is used to collect time domain data of each key node, and the time evolution and spatial propagation fitting of the fault are performed based on the time domain data.

[0034] The time evolution and space propagation fitting are used to perform fault source inversion authentication, and the fault is located based on the fault source inversion authentication result.

[0035] In a feasible implementation, after establishing fault location, the following steps are included:

[0036] After the fault level is established according to the fault location, a multi-level decision response scheme matching result is established, and the multi-level decision response scheme matching result includes a fault location response decision and a fault path evolution decision.

[0037] A fault is reported based on the matching result between the fault level and the multi-level decision response scheme.

[0038] In a feasible implementation, establishing fault location further includes:

[0039] Perform a prediction of the development of fault sources and establish an additional safety level based on the development prediction results.

[0040] A fault localization is established using the additional safety level and the fault level of the fault source.

[0041] In a second aspect, the present invention further provides a ground loop resistance detection and fault location system integrating edge computing, wherein the ground loop resistance detection and fault location system integrating edge computing includes:

[0042] The node deployment module is used to perform key node analysis of the ground loop, deploy edge computing units based on the key node analysis results, and configure monitoring sensors mapped to the edge computing units. The monitoring sensors include a high-frequency transient traveling wave detector, a broadband impedance spectrum measurement sensor, and a synchronous clock unit.

[0043] A field reconstruction module is used to perform power grid device monitoring using the monitoring sensor when the power grid device is operating normally. After aligning the time of the monitoring data set based on the synchronous clock unit, the edge computing unit interpolates and reconstructs the continuous situation field of the ground loop. The continuous situation field is constructed based on the three-dimensional fault characteristics.

[0044] The fault source inversion and positioning module is used to use the monitoring sensor to monitor the power grid device in real time, and to perform fault source inversion based on local height gradient characteristics and extreme area characteristics through the real-time monitoring results and the continuous situation field to establish fault location.

[0045] The present invention discloses a ground loop resistance detection and fault location method and system that integrates edge computing, including: performing key node analysis on the ground loop, deploying edge computing units based on the analysis results, and configuring corresponding monitoring sensors, wherein the monitoring sensors include a high-frequency transient traveling wave detector, a broadband impedance spectrum measurement sensor, and a synchronous clock unit; during the normal operation of the power grid device, the monitoring sensors are used to collect operating data, and time synchronization alignment of the monitoring data set is achieved based on the synchronous clock unit, and then the edge computing unit performs interpolation reconstruction to generate a ground loop continuous situation field based on three-dimensional fault characteristics; the power grid device is monitored in real time by the monitoring sensors, and the real-time monitoring results are combined with the continuous situation field to perform fault source inversion analysis based on local height gradient characteristics and extreme area characteristics, thereby realizing fault location. The ground loop resistance detection and fault location method and system that integrates edge computing disclosed in the present invention solve the technical problems of large data synchronization deviation, heavy computing and transmission load, and insufficient fault location accuracy, and achieves the technical effects of reducing response delay, optimizing load distribution, and improving fault location accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 This is a flow chart of the ground loop resistance detection and fault location method integrated with edge computing in the present invention.

[0047] Figure 2 This is a structural diagram of the ground loop resistance detection and fault location system integrated with edge computing in the present invention.

[0048] Description of the accompanying drawings: node deployment module 11, field reconstruction module 12, fault source inversion and positioning module 13. DETAILED DESCRIPTION

[0049] The above technical solution will be described in detail below in conjunction with the accompanying drawings and specific implementation methods of the specification to better understand the above technical solution. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments of the present invention. It should be understood that the present invention is not limited to the example embodiments used only to explain the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention. In addition, it should be noted that, for the convenience of description, only the parts related to the present invention, rather than all, are shown in the drawings.

[0050] Example 1, as Figure 1 This is a flow chart of a method for detecting and locating ground loop resistance and faults by integrating edge computing according to the present invention, wherein the method for detecting and locating ground loop resistance and faults by integrating edge computing includes:

[0051] S100: Perform key node analysis of the ground loop, deploy an edge computing unit based on the key node analysis results, and configure monitoring sensors mapped to the edge computing unit. The monitoring sensors include a high-frequency transient traveling wave detector, a broadband impedance spectrum measurement sensor, and a synchronous clock unit.

[0052] Specifically, based on the ground loop topology, node electrical characteristics, and historical operating data, nodes in the power grid ground loop with important electrical characteristics and a significant impact on fault propagation, known as critical nodes, are identified and evaluated. These critical nodes are typically located at key locations in the power grid topology, such as branch points, connection points, or locations where electrical parameters vary significantly. Critical node analysis can be performed using techniques such as electrical network analysis, sensitivity analysis, and fault simulation.

[0053] Specifically, edge computing units are deployed at identified key node locations. Each edge computing unit has the capabilities of local data collection, preliminary analysis, anomaly detection, data preprocessing and intelligent decision support, thereby enabling local rapid response, improving the real-time and accuracy of detection, and reducing the burden on the central server.

[0054] Specifically, edge computing units are computing resources located close to data sources or end devices, used to process local data and reduce reliance on central servers. Monitoring sensors are devices used to collect grid operating status data. These include high-frequency transient traveling wave detectors (for capturing high-frequency transient signals), broadband impedance spectroscopy sensors (for measuring impedance changes across a wide frequency band), and synchronous clock units (for ensuring time synchronization of data collection).

[0055] The above steps, through key node analysis, can accurately deploy monitoring resources and improve the coverage and efficiency of the detection system. At the same time, through the local processing capabilities of the edge computing unit, rapid response of fault detection and preliminary positioning can be achieved, reducing latency.

[0056] In some embodiments, performing a key node analysis of a ground loop and deploying an edge computing unit based on the key node analysis results includes:

[0057] Analyze and construct all arrangable nodes for grounding loops; after configuring fused multi-dimensional indicators, use the fused multi-dimensional indicators to evaluate the node importance of all arrangable nodes. The fused multi-dimensional indicators include electrical contribution index, propagation sensitivity index, spectrum heterogeneity index, time domain evolution coupling index, and fault evolution potential index; use the node importance evaluation results to perform proportional screening, perform node distribution compensation based on the proportional screening results, establish key nodes, and deploy edge computing units at the key nodes.

[0058] Specifically, based on the physical topology of the ground loop, all nodes where monitoring equipment can be installed are identified, including grounding down conductor nodes, grounding electrode nodes, and equipment grounding terminal nodes, among other nodes that can form a ground loop. For each deployable node, the required electrical and fault data are collected, and five fused multidimensional indicators are calculated or evaluated accordingly. Finally, each node is assigned an importance score based on a multi-indicator fusion algorithm (such as a weighted comprehensive score or machine learning classification).

[0059] Specifically, the integrated multidimensional indicators include:

[0060] Electrical contribution index: Indicates the degree of influence of a node on the overall electrical characteristics of the ground loop (such as total resistance and ground potential distribution). It is calculated by calculating the sensitivity of changes in node resistance to changes in global electrical parameters.

[0061] Propagation sensitivity index: Indicates the sensitivity of the node to the propagation path and intensity of the fault signal (such as transient traveling waves), that is, reflects the criticality of the node in the fault energy propagation link.

[0062] Spectral heterogeneity index: Indicates the degree of difference in the spectral characteristics of the collected signals at the node, and is used to measure the node's ability to distinguish abnormal signals in different frequency bands.

[0063] Time domain evolution coupling index: It indicates the dynamic correlation strength between a node and other nodes during the time domain signal evolution process, that is, it reflects the synchronization and coordination of nodes during the fault evolution process.

[0064] Fault evolution potential index: Indicates the possibility of a node serving as a starting point or relay point in the development of a potential fault, which can be obtained through historical fault data modeling or simulation.

[0065] Furthermore, proportional screening is performed based on the node importance evaluation results, that is, a certain proportion of the most important nodes are selected, such as setting the selection of nodes with the top 20% to 30% of the score, so as to retain high-importance nodes and eliminate low-importance nodes, and preliminarily form a candidate set of key nodes.

[0066] Finally, the spatial distribution of the initially screened nodes is analyzed to identify potential monitoring blind spots or areas with sparse nodes. Accordingly, a certain number of additional nodes are deployed or the node distribution is adjusted to ensure good spatial coverage and redundancy of key nodes. Furthermore, the finalized key nodes are used as the deployment locations for edge computing units. Each key node is equipped with an edge computing unit and its corresponding monitoring sensors (such as high-frequency transient traveling wave detectors, broadband impedance spectroscopy measurement sensors, and synchronous clock units), forming a distributed, intelligent ground loop monitoring and fault location network.

[0067] S200: When the power grid device is operating normally, the monitoring sensor is used to perform power grid device monitoring. After the monitoring data set is time-synchronized and aligned based on the synchronous clock unit, the continuous situation field of the ground loop is reconstructed through interpolation by the edge computing unit. The continuous situation field is constructed based on the three-dimensional fault characteristics.

[0068] Specifically, the continuous state field refers to a continuous distribution diagram of the electrical state of the ground loop (such as potential, impedance, and transient characteristics) in space and time, formed by interpolating and reconstructing data from discrete monitoring nodes during the normal operation of the power grid device; this continuous state field can dynamically reflect the evolution of the health status of the ground loop.

[0069] Specifically, three-dimensional fault characteristics refer to the three main characteristic dimensions used to describe the ground loop status, including frequency domain characteristics (such as impedance spectrum changes and harmonic components), time domain characteristics (such as transient waveforms and impulse responses), and spatial distribution characteristics (such as node potential gradients and energy distribution). Through these three-dimensional fault characteristics, it is possible to comprehensively describe the multi-dimensional information of the abnormal evolution of the ground loop, providing sufficient data for subsequent steps.

[0070] In some embodiments, reconstructing the continuous situation field of the ground loop by interpolation of the edge computing unit includes:

[0071] After using the synchronous clock unit to synchronize the monitoring data set, the time-aligned monitoring data set is sent to the corresponding edge computing unit. After the edge computing units form a computing local area network:

[0072] Perform spatial propagation feature extraction of high-frequency transient traveling waves to establish spatial propagation features, which include inter-node propagation delay features, attenuation coefficient features, and local reflection features; perform spectral response feature extraction of broadband impedance spectra to establish spectral response features, which include amplitude-frequency variation features, phase-frequency variation features, and frequency segmentation feature vectors; perform time-domain evolution feature extraction on monitoring data sets of multiple key nodes to establish time-domain evolution features, which include synchronous evolution features, node association features, and mutation features; construct three-dimensional fault features using the spatial propagation features, spectral response features, and time-domain evolution features, and reconstruct the continuous state field of the grounding loop after interpolating the three-dimensional fault features.

[0073] Specifically, first, after completing the time synchronization alignment of the monitoring data set of each monitoring node using the synchronous clock unit, the time-aligned monitoring data set is sent to the corresponding edge computing unit, where multiple edge computing units are interconnected through a local area network (LAN) to form a distributed collaborative computing environment.

[0074] Specifically, the spatial propagation characteristics are obtained based on high-frequency transient traveling wave signals and are used to describe the spatiotemporal characteristics of the signal during propagation between nodes, including: inter-node propagation delay characteristics (the time delay required for the signal to be transmitted from one node to another), attenuation coefficient characteristics (the degree of energy attenuation of the signal during propagation, such as the percentage of attenuation or the decibel value of attenuation), and local reflection characteristics (the phenomenon and characteristics of signal reflection at discontinuities in nodes or lines).

[0075] Specifically, the spectrum response characteristics are obtained based on broadband impedance spectrum measurement and are used to describe the impedance change law of the node at different frequencies. The spectrum response characteristics include: amplitude-frequency change characteristics (the characteristics of impedance amplitude changing with frequency), phase-frequency change characteristics (the characteristics of impedance phase changing with frequency) and frequency segmentation feature vectors (the frequency interval is divided into segments to extract features and form a multi-dimensional vector description to describe the response characteristics of the node in different frequency bands).

[0076] Specifically, the time-domain evolution features are acquired based on the time series data of multiple key nodes and are used to describe the dynamic characteristics of the ground loop status over time. These include: synchronous evolution features (the synchronization of signal changes at different nodes, such as the synchronization coefficient), node correlation features (the correlation of signal changes between nodes, such as the mutual correlation coefficient), and mutation features (sudden changes or anomalies in node signals, such as mutation point detection).

[0077] Specifically, a multidimensional feature body composed of spatial propagation characteristics, spectral response characteristics, and time domain evolution characteristics, namely the three-dimensional fault characteristics, is used to comprehensively describe the state characteristics of the ground loop in the three dimensions of space, frequency, and time.

[0078] Furthermore, the three-dimensional fault features are input into the edge computing unit, and a spatial interpolation algorithm (such as Kriging interpolation, radial basis function interpolation, sparse reconstruction algorithm, etc.) is used to reconstruct the continuous situation field within the ground loop. Optionally, the three-dimensional fault features (frequency domain, time domain, spatial distribution characteristics) are integrated during the interpolation process to improve the physical rationality and abnormal sensitivity of the reconstructed situation field.

[0079] The continuous state field reconstructed in the above steps can truly reflect the spatial distribution and temporal changes of the electrical state in the grounding loop. Then, through dynamic monitoring of the continuous state field, hidden dangers such as abnormal growth of grounding resistance and incipient small-scale faults can be identified in advance.

[0080] In some embodiments, reconstructing the continuous situation field of the ground loop after interpolating the three-dimensional fault signature includes:

[0081] Based on the dependency relationships of key nodes, connection edges are established at all key nodes; connection edge weights are configured based on the spatial distance and feature difference of the connection edges; after establishing strength and direction constraints for feature propagation transition based on the connection edge weights, feature interpolation is performed through traversing the three-dimensional fault features to reconstruct the continuous situation field of the ground loop.

[0082] Specifically, the interpolation process can be guided by the graph structure, so that the interpolation not only considers the spatial position, but also the feature similarity and propagation direction, including: building edges based on the dependency relationship between key nodes, configuring the weights of connecting edges (based on spatial distance and feature difference), establishing strength constraints and direction constraints for feature propagation transitions, and performing feature interpolation based on this.

[0083] Specifically, key node dependencies refer to logical dependencies between different monitoring nodes in a ground loop based on physical connections, geographic proximity, or electrical characteristic correlations. Edges are connecting lines established between key nodes based on these dependencies, representing the feature propagation paths between nodes. Edge weights represent the importance or propagation capability of the edge relationship and are determined based on spatial distance and feature diversity.

[0084] Illustratively, the spatial distance includes the physical or electrical distance between nodes, and the feature difference involves the similarity of three-dimensional fault features between nodes, such as Euclidean distance, cosine similarity, etc.

[0085] Specifically, the feature propagation transition constraint refers to the control of the strength of feature propagation from one node to another according to the connection edge weight during the feature interpolation process; the direction constraint is the priority direction or restricted direction of feature propagation set according to the feature change trend between nodes during the feature interpolation process.

[0086] Specifically, the above steps are implemented by first establishing edges between all key nodes based on physical connections, geographic proximity, or electrical characteristic correlations. Each edge is weighted based on spatial distance and feature diversity. For example, the greater the distance, the smaller the weight, and the greater the difference, the smaller the weight. Optionally, edge weights can be calculated using a weighted function that combines spatial distance and feature diversity.

[0087] Specifically, based on the connection edge weights, the strength control rules of feature propagation are established, that is, the priority direction or prohibited direction of feature propagation is determined to form a graph structure with weighted directed edges. Then, on the graph structure, according to the connection edge weights and direction constraints, weighted averaging, graph convolution or random walk-based interpolation methods are used to expand the discrete three-dimensional fault features into a continuous distribution, thereby completing the reconstruction of the continuous situation field of the grounding loop.

[0088] The aforementioned method, by introducing edge weights based on spatial and feature differences, ensures that feature propagation adheres to actual physical laws, avoiding meaningless leapfrog interpolation. Furthermore, directional constraints ensure that fault features propagate along the actual propagation path (e.g., current flow), improving fault location accuracy. The weighted directed graph structure effectively suppresses the interference of abnormal nodes on the overall interpolation results, improving fault tolerance to local anomalies.

[0089] S300: Using the monitoring sensor to monitor the power grid device in real time, and performing fault source inversion based on local height gradient characteristics and extreme value area characteristics through the real-time monitoring results and the continuous situation field to establish fault location.

[0090] Specifically, after establishing the continuous situation field, the ground loop status of the power grid device is continuously monitored and collected in real time through monitoring sensors, including high-frequency transient traveling wave detectors, broadband impedance spectrum measurement sensors, and synchronous clock units to collect high-frequency signals, impedance signals, and time synchronization. This generates real-time monitoring results, which together with the continuous situation field serve as the basis for subsequent fault location.

[0091] In some embodiments, performing fault source inversion based on local height gradient characteristics and extreme area characteristics through real-time monitoring results and the continuous situation field to establish fault location includes:

[0092] After the real-time monitoring results are time-synchronized and aligned, the time-synchronized and aligned real-time monitoring results are superimposed on the continuous situation field according to time slices to form a dynamic situation incremental map; the local gradient is calculated on the dynamic situation incremental map, and the gradient modulus is established; the high-gradient area is screened using the gradient modulus to establish a local high-gradient feature; the local maximum point and the local minimum point are identified according to the local gradient using a local neighborhood search to establish an extreme value area feature; the fault source is inverted using the local high gradient feature and the extreme value area feature.

[0093] Specifically, real-time monitoring results refer to data collected by monitoring sensors during grid operation, reflecting the current state changes of the grid. The dynamic incremental situation graph, created by superimposing real-time monitoring results onto the continuous situation field, reflects the dynamic changes in fault characteristics and displays the incremental changes in fault characteristics over time.

[0094] Specifically, the local gradient is the rate of change of the fault feature between a certain point in the dynamic situation increment graph and its surrounding points, reflecting the spatial variation trend of the fault feature. The gradient modulus corresponds to the magnitude of the local gradient and is used to measure the intensity of the change in the fault feature. The local high gradient feature corresponds to the area with a higher gradient modulus in the dynamic situation increment graph, indicating the location where the fault feature has changed significantly. The extreme value region feature is the local maximum and local minimum points identified through local neighborhood search, which can be considered as a significant sign of the fault feature, that is, the extreme value region feature may be related to the fault source.

[0095] Specifically, first, the real-time monitoring results collected by each monitoring node are time-synchronized and aligned. For example, the synchronized real-time monitoring results are superimposed on the current continuous situation field according to time slices to form a dynamic situation increment map, which reflects the local change trend after the fault occurs. Then, numerical differentiation or finite difference methods are used to calculate the local gradient and gradient modulus (i.e., the modulus length of the gradient vector) on the dynamic situation increment map. Next, based on a preset threshold, the area where the gradient modulus exceeds the threshold is screened as a local high-gradient area, and the local high-gradient feature is extracted. Then, based on the local neighborhood search algorithm (such as sliding window, maximum and minimum value detection) on the dynamic situation increment map, the local maximum and local minimum points are identified, and the extreme value area feature is established accordingly. Optionally, the extreme value area feature is used to record the position, amplitude, and other information of the extreme value point.

[0096] Furthermore, a comprehensive analysis is conducted on the distribution of local high-gradient areas and extreme points, and the fault source location is inferred based on the degree of feature aggregation, amplitude, and spatial distribution pattern, and inversion optimization is performed to further accurately locate the fault location.

[0097] Through the above process, real-time monitoring results are time-synchronized with the continuous situation field to form a dynamic situation increment map, which can capture the dynamic changes of fault characteristics in real time. Calculating local gradients and establishing gradient moduli can effectively identify areas with significant changes in fault characteristics. Furthermore, fault source inversion based on local high-gradient features and extreme value area features can significantly improve the accuracy and reliability of fault location, facilitate rapid response to fault changes, and provide strong support for power grid fault handling and maintenance.

[0098] In some implementations, after establishing the gradient modulus, the method further includes:

[0099] Obtain a fixed decision threshold for matching the current listening scene; calculate the mean and standard deviation of the global gradient field based on the gradient modulus, and construct a sensitivity adjustment factor using the mean and standard deviation calculation results; use the sensitivity adjustment factor to perform focus enhancement on the fixed decision threshold for matching, and then use the enhanced fixed decision threshold for matching to screen high gradient areas.

[0100] Specifically, the matching fixed threshold is a pre-set fixed threshold based on the current monitoring scenario (e.g., grid operating status, fault type, etc.). It is used to determine whether the gradient modulus is large enough to indicate a significant change in the fault signature. The global gradient field includes all gradient moduli calculated in the entire dynamic situation increment map, reflecting the distribution of the intensity of fault signature changes across the entire monitoring area.

[0101] Specifically, by calculating the statistical characteristics of the global gradient field (such as the mean and standard deviation), the corresponding sensitivity adjustment factor can be obtained. The sensitivity adjustment factor is used to adjust the sensitivity of the decision threshold to adapt to different fault feature intensity distributions. For example, the sensitivity adjustment factor is constructed based on the following formula:

[0102] ;

[0103] Among them, k is the sensitivity adjustment factor, γ is the adjustment coefficient used to control the sensitivity enhancement amplitude, σ and μ are the mean and standard deviation of the global gradient field, respectively.

[0104] Furthermore, the matching fixed decision threshold is adjusted by the sensitivity adjustment factor to obtain the enhanced matching fixed decision threshold to extract the local high gradient area, forming a more accurate and sensitive local height gradient feature.

[0105] The above process can adaptively adjust the high-gradient recognition standard according to different systems and environments by matching the fixed judgment threshold with the sensitivity adjustment factor. When the fault signal is weak or the background noise is large, it can effectively improve the recognition rate of the high-gradient area, reduce the missed detection rate, and avoid misjudgment or missed judgment caused by unreasonable fixed threshold setting.

[0106] In some implementations, performing fault source inversion based on the local height gradient feature and the extreme value region feature includes:

[0107] After locating the fault source, the synchronous clock unit is used to collect time domain data of each key node, and the time evolution and spatial propagation fitting of the fault are performed based on the time domain data; the time evolution and spatial propagation fitting are used to perform fault source inversion authentication, and the fault is located based on the fault source inversion authentication result.

[0108] Specifically, after initially locating the fault source, time-domain data from key nodes is collected (time alignment is ensured using synchronized clock units). Based on this time-domain data, the temporal evolution and spatial propagation of the fault are fitted. The fitting results are then used to verify the fault source and determine the fault location. This process effectively adds a step for temporal and spatial consistency verification, which improves the accuracy and reliability of fault location and avoids potential misjudgments based solely on spatial features (gradients, extreme values).

[0109] Specifically, first, based on the local height gradient characteristics and extreme area characteristics, the preliminary fault source inversion and positioning are completed; then, the synchronous clock unit is used to collect high-precision time domain data of each key node during the fault occurrence period of the preliminary fault source inversion and positioning, including waveform information of voltage, current, impedance, etc. that changes with time; then, based on the time domain data of each key node, the fault characteristics that change with time are analyzed, such as the rise, fall, oscillation and other dynamic characteristics of the fault voltage and current signals, and a time evolution curve of the fault characteristics is established (such as using least squares fitting, spatiotemporal regression analysis and other methods) to reflect the dynamic process of the fault spreading from the source point to the outside, including the propagation path, propagation speed and propagation direction of the fault signal.

[0110] Furthermore, the rationality of the preliminary fault source location is verified through the time evolution fitting and spatial propagation fitting results: the spatial propagation fitting results are traced back to the starting point, and it is determined whether the starting point is consistent with the preliminary fault source location result; if consistent, the preliminary location is confirmed to be correct; if inconsistent, the fault source position is corrected according to the fitting results.

[0111] By introducing spatiotemporal fitting authentication after preliminary positioning, the above steps can significantly reduce the error caused by simple spatial feature positioning and eliminate the impact of occasional noise and local anomalies on the positioning results.

[0112] In some embodiments, after establishing fault location, the steps include:

[0113] After establishing the fault level according to the fault location, a multi-level decision response scheme matching result is established, and the multi-level decision response scheme matching result includes a fault location response decision and a fault path evolution decision; and a fault is reported based on the fault level and the multi-level decision response scheme matching result.

[0114] Optionally, based on fault location, the fault is first classified based on the fault location results, combined with factors such as fault severity, impact scope, and evolution trend, to establish a fault level. Then, based on the fault level, a multi-level decision-making response solution is matched from a pre-set response solution library. Exemplary multi-level decision-making response solutions include: fault location response decisions (such as isolation, switching, and load shedding) and fault path evolution decisions (such as predicting fault expansion trends and alerting surrounding nodes). Finally, the fault is reported based on the fault level and multi-level response solution.

[0115] Exemplarily, the fault reporting information includes: fault source location, fault level, recommended response measures, warning expansion path, etc.

[0116] Through the above process, fault detection is essentially expanded from positioning to intelligent response and disposal, providing a closed-loop capability of fault perception-decision-response, which in turn helps relevant personnel to quickly respond to and handle faults.

[0117] In some implementations, establishing fault location further includes:

[0118] A development prediction of the fault source is performed, and an additional safety level is established based on the development prediction result; and a fault location is established using the additional safety level and the fault level of the fault source.

[0119] Furthermore, a corresponding fault prediction model can be constructed through fitting analysis models, physical modeling, time series prediction, and machine learning regression models, thereby predicting the development of the fault source and obtaining corresponding development prediction results. These development prediction results include the fault severity, impact range, evolution trend, etc. after a preset time, and are associated with the corresponding additional safety level. Then, using strategies such as weighting, taking the maximum value, and priority enhancement, the current fault level of the fault source and the additional safety level are combined to obtain a comprehensive fault level. Based on the comprehensive fault level, the final fault location result is determined through the same threshold screening and extreme value selection methods mentioned above.

[0120] The above process is not only based on current detection results, but also introduces considerations of future development trends. This helps to perceive potential threats in advance, thereby enhancing the safety margin of fault handling, and helping fault management move from passive response to active prevention, ultimately reducing the overall operational risk of the system.

[0121] In summary, the ground loop resistance detection and fault location method integrated with edge computing provided by the present invention has the following technical effects:

[0122] By analyzing the key nodes of the ground loop, edge computing units are deployed based on the analysis results, and corresponding monitoring sensors are configured. The monitoring sensors include a high-frequency transient traveling wave detector, a broadband impedance spectrum measurement sensor, and a synchronous clock unit. During the normal operation of the power grid device, the monitoring sensors are used to collect operating data, and the time synchronization of the monitoring data set is achieved based on the synchronous clock unit. The edge computing unit then performs interpolation and reconstruction to generate a continuous ground loop situation field based on three-dimensional fault characteristics. The power grid device is monitored in real time by the monitoring sensors. The real-time monitoring results are combined with the continuous situation field to perform fault source inversion analysis based on local height gradient characteristics and extreme area characteristics, thereby realizing fault location, thereby achieving the technical effects of reducing response delay, optimizing load distribution, and improving fault location accuracy.

[0123] Example 2, as Figure 2 This is a schematic diagram of the structure of the ground loop resistance detection and fault location system integrated with edge computing in the present invention. For example, Figure 1 The flow chart of the ground loop resistance detection and fault location method integrating edge computing in the present invention can be shown as follows: Figure 2The structure shown is implemented.

[0124] Based on the same concept as the ground loop resistance detection and fault location method integrated with edge computing in the above embodiment, the present invention also provides a ground loop resistance detection and fault location system integrated with edge computing, including:

[0125] The node deployment module 11 is used to perform key node analysis of the ground loop, deploy edge computing units based on the key node analysis results, and configure monitoring sensors mapped to the edge computing units. The monitoring sensors include high-frequency transient traveling wave detectors, broadband impedance spectrum measurement sensors, and synchronous clock units.

[0126] The field reconstruction module 12 is used to perform power grid device monitoring using the monitoring sensor when the power grid device is operating normally. After aligning the time of the monitoring data set based on the synchronous clock unit, the continuous situation field of the ground loop is reconstructed through interpolation by the edge computing unit. The continuous situation field is constructed based on the three-dimensional fault characteristics.

[0127] The fault source inversion and positioning module 13 is used to use the monitoring sensor to monitor the power grid device in real time, and perform fault source inversion based on local height gradient characteristics and extreme value area characteristics through the real-time monitoring results and the continuous situation field to establish fault location.

[0128] In some embodiments, the node deployment module 11 includes:

[0129] The routable node analysis and construction unit is used to analyze and construct all routable nodes of the ground loop.

[0130] The node importance evaluation unit is used to configure the fused multidimensional indicators and then use the fused multidimensional indicators to evaluate the node importance of all deployable nodes. The fused multidimensional indicators include electrical contribution index, propagation sensitivity index, spectrum heterogeneity index, time domain evolution coupling index, and fault evolution potential index.

[0131] The key node establishment and edge computing unit deployment unit is used to use the node importance evaluation results to perform proportional screening, perform node distribution compensation based on the proportional screening results, establish key nodes, and deploy edge computing units at the key nodes.

[0132] In some embodiments, the field reconstruction module 12 includes:

[0133] The monitoring data synchronization and edge computing unit transmission unit is used to synchronize the monitoring data set time using the synchronization clock unit, and then send the time-aligned monitoring data set to the corresponding edge computing unit to form a computing local area network with the edge computing units.

[0134] The high-frequency transient traveling wave spatial propagation feature extraction unit is used to perform spatial propagation feature extraction of the high-frequency transient traveling wave and establish spatial propagation features, wherein the spatial propagation features include inter-node propagation delay features, attenuation coefficient features, and local reflection features.

[0135] The broadband impedance spectrum spectrum response feature extraction unit is used to extract the broadband impedance spectrum spectrum response features and establish spectrum response features. The spectrum response features include amplitude-frequency variation features, phase-frequency variation features, and frequency segmentation feature vectors.

[0136] The time domain evolution feature extraction and establishment unit is used to extract time domain evolution features from the monitoring data sets of multiple key nodes and establish time domain evolution features. The time domain evolution features include synchronous evolution features, node association features, and mutation features.

[0137] The three-dimensional fault feature building and continuous situation field reconstruction unit is used to build a three-dimensional fault feature based on the spatial propagation feature, spectrum response feature, and time domain evolution feature, and reconstruct the continuous situation field of the ground loop after interpolating the three-dimensional fault feature.

[0138] In some implementations, the three-dimensional fault feature building and continuous situation field reconstruction unit in the field reconstruction module 12 includes:

[0139] The key node connection edge establishing unit is used to establish connection edges at all key nodes based on the dependency relationship of the key nodes.

[0140] The connection edge weight configuration unit is used to configure the connection edge weight according to the spatial distance and feature difference of the connection edge.

[0141] The feature propagation constraint and continuous situation field reconstruction unit is used to establish the strength constraint and direction constraint of the feature propagation transition based on the connection edge weight, and then traverse the three-dimensional fault features to perform feature interpolation to reconstruct the continuous situation field of the ground loop.

[0142] In some embodiments, the fault source inversion and location module 13 includes:

[0143] The real-time monitoring result time synchronization and superposition unit is used to synchronize the time of the real-time monitoring results and then superpose the time-synchronized real-time monitoring results onto the continuous situation field according to time slices to form a dynamic situation increment map.

[0144] The local gradient calculation and gradient modulus establishment unit is used to calculate the local gradient on the dynamic situation increment map and establish the gradient modulus.

[0145] The local high gradient feature establishing unit is used to use the gradient modulus to screen the high gradient area and establish the local high gradient feature.

[0146] The extreme value region feature establishment unit is used to identify local maximum points and local minimum points according to local gradients using local neighborhood search to establish extreme value region features.

[0147] The fault source inversion unit is used to perform fault source inversion based on the local high gradient characteristics and extreme value area characteristics.

[0148] In some implementations, the local gradient calculation and gradient modulus establishment unit in the fault source inversion and location module 13 includes:

[0149] The fixed decision threshold acquisition subunit is used to acquire a fixed decision threshold that matches the current monitoring scene.

[0150] The global gradient field statistics and sensitivity adjustment factor construction subunit is used to calculate the mean and standard deviation of the global gradient field according to the gradient modulus value, and construct the sensitivity adjustment factor using the mean and standard deviation calculation results.

[0151] The high gradient region screening subunit is used to use the sensitivity adjustment factor to perform matching fixed decision threshold focus enhancement, and then perform high gradient region screening with the matching fixed decision threshold of the focus enhancement.

[0152] In some implementations, the fault source inversion unit in the fault source inversion and location module 13 includes:

[0153] The time domain data collection unit after fault source location is used to collect time domain data of each key node using the synchronous clock unit after the fault source is located.

[0154] The fault time evolution and spatial propagation fitting unit is used to perform time evolution and spatial propagation fitting of the fault based on the time domain data.

[0155] The fault source inversion authentication and positioning unit is used to perform fault source inversion authentication by using the time evolution and space propagation fitting, and to perform fault positioning based on the fault source inversion authentication result.

[0156] In some embodiments, the fault source inversion and location module 13 further includes:

[0157] The fault level and decision response scheme matching unit is used to establish a multi-level decision response scheme matching result after establishing a fault level according to the fault location. The multi-level decision response scheme matching result includes a fault location response decision and a fault path evolution decision.

[0158] A fault reporting unit is used to report a fault based on a matching result between the fault level and the multi-level decision response solution.

[0159] In some embodiments, the fault source inversion and location module 13 further includes:

[0160] The fault development prediction unit is used to perform development prediction of the fault source and establish an additional safety level based on the development prediction result.

[0161] The additional fault location unit is configured to establish a fault location using the additional safety level and the fault level of the fault source.

[0162] It should be understood that the embodiments mentioned in this specification focus on their differences from other embodiments. The specific embodiments in the aforementioned embodiment one are also applicable to the ground loop resistance detection and fault location system integrated with edge computing described in embodiment two. For the sake of brevity of the specification, no further elaboration is given here.

[0163] It should be understood that the embodiments disclosed in the present invention and the above description can enable those skilled in the art to use the present invention to implement the present invention. At the same time, the present invention is not limited to the embodiments mentioned above. It should be understood that those skilled in the art can still modify the technical solutions described in the above embodiments or replace some of the technical features therein with equivalents; and such modifications or replacements do not deviate from the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention and are all included in the scope of protection of the present invention.

Claims

1. A ground loop resistance detection and fault location method integrating edge computing is characterized by: The method comprises: Perform key node analysis of the ground loop, deploy edge computing units based on the key node analysis results, and configure monitoring sensors mapped to the edge computing units, including a high-frequency transient traveling wave detector, a broadband impedance spectrum measurement sensor, and a synchronous clock unit; Under the condition that the power grid device is operating normally, the monitoring sensor is used to perform power grid device monitoring, and after the monitoring data set is time-synchronized and aligned based on the synchronous clock unit, the continuous situation field of the ground loop is reconstructed through interpolation by the edge computing unit. The continuous situation field is constructed based on the three-dimensional fault characteristics; Using the monitoring sensor to monitor the power grid device in real time, and performing fault source inversion based on local height gradient characteristics and extreme value area characteristics through the real-time monitoring results and the continuous situation field to establish fault location; The method of reconstructing the continuous situation field of the ground loop by interpolation of the edge computing unit includes: After using the synchronous clock unit to synchronize the monitoring data set, the time-aligned monitoring data set is sent to the corresponding edge computing unit. After the edge computing units form a computing local area network: Perform spatial propagation feature extraction of high-frequency transient traveling waves and establish spatial propagation features, wherein the spatial propagation features include inter-node propagation delay features, attenuation coefficient features, and local reflection features; Execute spectrum response feature extraction of broadband impedance spectrum and establish spectrum response features, wherein the spectrum response features include amplitude-frequency variation features, phase-frequency variation features, and frequency segmentation feature vectors; Extracting time-domain evolution features from monitoring data sets of multiple key nodes to establish time-domain evolution features, including synchronous evolution features, node association features, and mutation features; The three-dimensional fault feature is constructed by using the spatial propagation feature, the spectrum response feature, and the time domain evolution feature, and after interpolating the three-dimensional fault feature, a continuous state field of the ground loop is reconstructed; The method of performing fault source inversion based on local height gradient characteristics and extreme area characteristics through the real-time monitoring results and the continuous situation field to establish fault location includes: After aligning the time of the real-time monitoring results, superimposing the time-synchronized real-time monitoring results onto the continuous situation field according to time slices to form a dynamic situation increment map; Calculating local gradients on the dynamic situation increment graph and establishing gradient modulus values; Using the gradient modulus to screen high gradient areas and establish local high gradient features; Use local neighborhood search to identify local maximum and local minimum points based on local gradients to establish extreme area features; The fault source is inverted using the local height gradient characteristics and extreme value area characteristics.

2. The ground loop resistance detection and fault location method integrated with edge computing according to claim 1, characterized in that: After interpolating the three-dimensional fault features, reconstructing the continuous situation field of the ground loop includes: Establish connection edges at all key nodes based on the dependency relationships of key nodes; Configure the edge weights by the spatial distance and feature difference of the edges; After establishing the strength constraint and direction constraint of feature propagation transition based on the connection edge weight, the three-dimensional fault features are traversed to perform feature interpolation to reconstruct the continuous situation field of the ground loop.

3. The ground loop resistance detection and fault location method integrated with edge computing according to claim 1, wherein: The performing of the key node analysis of the ground loop and deploying the edge computing unit based on the key node analysis results includes: Analyze all routable nodes to construct a ground loop; After configuring the fusion multi-dimensional indicators, the node importance evaluation of all deployable nodes is performed using the fusion multi-dimensional indicators, which include the electrical contribution index, propagation sensitivity index, spectrum heterogeneity index, time domain evolution coupling index, and fault evolution potential index; The node importance evaluation results are used for proportional screening. After node distribution compensation is performed based on the proportional screening results, key nodes are established and edge computing units are deployed at the key nodes.

4. The ground loop resistance detection and fault location method integrated with edge computing according to claim 1, wherein: After the gradient modulus value is established, the method includes: Get the fixed threshold for matching the current monitoring scene; Calculating the mean and standard deviation of the global gradient field according to the gradient modulus, and constructing a sensitivity adjustment factor using the calculation results of the mean and standard deviation; After the sensitivity adjustment factor is used to perform focus enhancement on the matching fixed decision threshold, high gradient regions are screened using the focus enhanced matching fixed decision threshold.

5. The ground loop resistance detection and fault location method integrated with edge computing according to claim 1, wherein: The performing of fault source inversion based on the local height gradient characteristics and extreme value area characteristics includes: After locating the fault source, using the synchronous clock unit to collect time domain data of each key node, and performing temporal evolution and spatial propagation fitting of the fault based on the time domain data; The time evolution and space propagation fitting are used to perform fault source inversion authentication, and the fault is located based on the fault source inversion authentication result.

6. The ground loop resistance detection and fault location method integrated with edge computing according to claim 1, wherein: After the fault location is established, the following steps are included: After establishing the fault level according to the fault location, establishing a multi-level decision response solution matching result, the multi-level decision response solution matching result includes a fault location response decision and a fault path evolution decision; A fault is reported based on the matching result between the fault level and the multi-level decision response scheme.

7. The ground loop resistance detection and fault location method integrated with edge computing according to claim 1, wherein: The establishing of fault location further includes: Perform development prediction of fault sources and establish additional safety levels based on the development prediction results; A fault localization is established using the additional safety level and the fault level of the fault source.

8. The ground loop resistance detection and fault location system integrated with edge computing is characterized by: A method for detecting and locating ground loop resistance and faults for implementing the fused edge computing described in any one of claims 1 to 7, comprising: A node deployment module is used to perform key node analysis of the ground loop, deploy edge computing units based on the key node analysis results, and configure monitoring sensors mapped to the edge computing units. The monitoring sensors include a high-frequency transient traveling wave detector, a broadband impedance spectrum measurement sensor, and a synchronous clock unit. a field reconstruction module, configured to, when the power grid device is operating normally, monitor the power grid device using the monitoring sensor, align the time of the monitoring data set based on the synchronization clock unit, and then reconstruct a continuous situation field of the ground loop through interpolation by the edge computing unit, wherein the continuous situation field is constructed based on the three-dimensional fault characteristics; The fault source inversion and positioning module is used to use the monitoring sensor to monitor the power grid device in real time, and to perform fault source inversion based on local height gradient characteristics and extreme area characteristics through the real-time monitoring results and the continuous situation field to establish fault location.

Images