Overhead transmission line hidden trouble fault monitoring and diagnosing method based on side end calculation

Abstract

The invention discloses an overhead transmission line hidden trouble fault monitoring and diagnosing method based on side end calculation, relates to the technical field of line fault detection, and solves the problems that in the prior art, edge nodes only serve as data forwarding channels instead of cooperative calculation units; and the system is caught in a rigid cycle of data cloud-central modeling-unified issuing-static execution. According to the method, homogeneous regions are divided through environment-density clustering, an exclusive lightweight model is trained, an electrical topological weighted graph is constructed to quantify a fault propagation coefficient between nodes, accurate wake-up rather than full-region blind start is realized, and energy consumption is reduced; moreover, a fault frequency dynamic scheduling algorithm is introduced to be combined with a topological coupling compensation and power consumption budget game, and response delay is compressed through an event-driven hierarchical wake-up chain and multi-node weighted voting diagnosis, so that a self-sensing, self-decision-making and self-optimizing closed-loop collaborative architecture is formed; the core defects of idle edge node capability, collaboration deficiency and low energy efficiency are systematically solved.

Description

Technical Field

[0001] This invention relates to the field of line fault detection technology, and in particular to a method for monitoring and diagnosing hidden faults in overhead transmission lines based on edge-end calculation. Background Technology

[0002] With the advancement of smart grid construction and the maturity of IoT technology, edge computing architecture has been widely applied in the monitoring of overhead transmission lines. Traditional centralized monitoring transmits massive amounts of sensor data to the cloud, facing bottlenecks such as communication bandwidth constraints, central server overload, and high fault response latency, making it difficult to meet the needs of real-time perception and rapid response to line conditions. Edge computing, by deploying edge intelligent devices at key nodes such as line towers, enables local data preprocessing and real-time decision-making, significantly reducing cloud dependence and improving response speed. However, overhead transmission lines are characterized by highly heterogeneous environments (significant climatic differences between mountainous areas, plains, and polluted coastal areas) and strong topological correlation (electrical faults can propagate along conductors). Existing edge applications mostly remain at a shallow level of "data acquisition—simple threshold judgment—anomaly reporting," lacking collaborative mechanisms between edge nodes, highlighting the contradiction between model universality and regional adaptability, and failing to establish a refined power consumption management mechanism under energy-constrained conditions, thus failing to fully unleash the potential of edge computing.

[0003] The fundamental shortcoming of existing technologies lies in the fact that edge nodes merely serve as data forwarding channels rather than collaborative computing units, causing the system to fall into a rigid cycle of "data uploading to the cloud—centralized modeling—unified distribution—static execution." Specifically, this manifests as follows: each node independently runs a homogeneous model, failing to adapt to regional micro-environmental differences, resulting in model generalization errors exceeding 15%; there is a lack of electrical topology awareness between nodes, meaning that an alarm in any node triggers a full-area wake-up, leading to over 60% of energy wasted; and the monitoring frequency remains fixed, resulting in excessive sampling and energy waste during low-risk periods, while response is sluggish during periods of widespread failures.

[0004] Therefore, a method for monitoring and diagnosing potential faults in overhead transmission lines based on edge computing is needed. Summary of the Invention

[0005] To address the problem in existing technologies where edge nodes merely serve as data forwarding channels rather than collaborative computing units, leading to a rigid cycle of "data uploading to the cloud—centralized modeling—unified distribution—static execution," this invention provides a method for monitoring and diagnosing potential faults in overhead transmission lines based on edge computing. This method divides the entire network into K highly cohesive, homogeneous regions by constructing an "environment-density" dual-constraint region partitioning mechanism. Each region trains its own lightweight model, reducing model adaptation errors. An electrical topology weighted graph is established to quantify the fault propagation coefficient between nodes, waking only affected nodes and reducing invalid wake-ups. A dynamic fault frequency scheduling algorithm is introduced, combined with topology coupling compensation and power budget game theory, to adaptively adjust the node sampling frequency according to risk. This reduces total system energy consumption while increasing the detection frequency in high-risk areas, ultimately forming a closed-loop architecture of "self-sensing, self-decision-making, and self-optimization," systematically solving the core defects of idle edge node capabilities and lack of collaboration. The specific technical solution is as follows: A method for monitoring and diagnosing potential faults in overhead transmission lines based on edge-end computation includes the following steps: S1: Based on environmental characteristics such as altitude and climate and node distribution density, the monitoring network is dynamically divided into several homogeneous regions with high data cohesion through composite similarity calculation, thus defining the optimal data boundary for subsequent model training; S2: Train multimodal fault identification models adapted to the local microenvironment using historical data from each region. Ensure that the model has both generalizability and retains regional specificity through global-local difference constraints, and output model performance indicators for subsequent evaluation. S3: Construct a directed weighted graph based on the line impedance matrix and spatial distance, calculate the fault propagation weight of each node to other nodes, and form a quantitative basis matrix for wake-up decision and diagnosis voting; S4: Calculate the failure rate of each region and introduce the coefficient of variation correction. Combine topology coupling compensation and total power consumption constraints to generate an adaptive model start frequency for each node, so as to achieve optimal energy efficiency scheduling for higher risks and more intensive monitoring. S5: When a node detects an anomaly, it immediately initiates a three-level wake-up chain of local-regional-cross-regional nodes, activates associated nodes based on the propagation weight in step S3, and completes the fault determination by merging the inference results of multiple nodes through weighted voting.

[0006] Preferably, step S1 is as follows: S102: Obtain the set of edge node locations P, the environmental feature matrix E, and the node communication radius R, and define the comprehensive distance between nodes i and j, as follows: In the formula, The distance is the geographic Euclidean distance; The Euclidean distance is standardized based on environmental characteristics. Let i be the set of the communicating neighbors of node i. Let j be the set of the communicating neighbors of node j. These are the weighting coefficients; S102: Perform density clustering partitioning.

[0007] Preferably, step 2 is as follows: S201: Construct a time series dataset from the historical data of region k, and train a multimodal fault identification model adapted to the local microenvironment using the historical data of each region. S202: Employing a federated transfer learning framework, enabling the local model M of region k... k minimize; S203: The cloud periodically evaluates the model performance in each region and decides whether to trigger model retraining based on the evaluation results.

[0008] Preferably, step S3 is as follows: S301: Construct a directed weighted graph G = (V, E, W); where vertices V are all edge nodes, edges E represent electrical connections (based on the line impedance matrix Z); and weights W represent the fault propagation coefficient from node i to j. In the formula, This is the attenuation coefficient (typically 0.01~0.05). Let i be the electrical neighbor of node i; S302: When node i triggers an early warning, the impact vector is: S303: Set the impact threshold θ, and obtain the set of nodes that need to be woken up based on the threshold: .

[0009] Preferably, step S4 is as follows: S401: Perform area fault frequency monitoring statistics;.

[0010] S402: Calculate the activation frequency of the fault model, the theoretical sampling frequency of node i in region k: In the formula, This indicates the minimum / maximum sampling frequency set by the system; This indicates the electrical influence of node i on its neighbors; Indicates the coupling strength coefficient; S403: Frequency correction based on energy consumption constraints, power consumption model of node i: in, Indicates upload flag, total power consumption constraint: The corrected frequency is determined using the Lagrange multiplier method. .

[0011] Preferably, step S5 is as follows: S501: Establish a hierarchical wake-up mechanism, as detailed below: If node i detects an anomaly locally, then: 1. L1 local wake-up: Node i immediately switches to f max, continuous T burst seconds; 2. L2 Region Wake-up: Broadcasts an alarm to all nodes within region k, affecting the set A of nodes. i according to: 3. L3 cross-region wake-up: If the fault frequency in region k is... λk > l critical, adjacent area Rk +1 enters alert mode, frequency increases by 30%; S502: Wake-up node set A i Upload features to the edge aggregation node and perform integrated diagnostics: in, Indicates the weight.

[0012] Preferred weight Calculated using the following formula: .

[0013] Preferably, step S6 is as follows: S601: Obtain the comprehensive risk score for region k: In the formula, Representation Model Since the last update, This represents the weights, and the sum of the three is 1; S602: According to Adjust the resource quota for region k.

[0014] A computer-readable storage medium includes a stored program, wherein, when the program is executed, it controls the device where the computer-readable storage medium is located to perform the edge computing-based overhead transmission line hidden fault monitoring and diagnosis method as described above.

[0015] A processor for running a program, wherein the program executes the edge-computation-based method for monitoring and diagnosing potential faults in overhead transmission lines as described above.

[0016] Compared with the prior art, the beneficial effects of the present invention are as follows: This invention overcomes the generalization error bottleneck of traditional homogeneous models by dividing homogeneous regions into homogeneous regions through environment-density clustering and training a dedicated lightweight model. It constructs an electrical topology weighted graph to quantify the fault propagation coefficient between nodes, achieving precise wake-up rather than blind start-up of the entire region, thus reducing ineffective energy consumption. It introduces a fault frequency dynamic scheduling algorithm combined with topology coupling compensation and power budget game theory, enabling the monitoring frequency to adaptively adjust with risk, reducing total system energy consumption while increasing the detection frequency of high-risk areas. Finally, through an event-driven hierarchical wake-up chain and multi-node weighted voting diagnosis, it compresses response latency, forming a self-aware, self-decision-making, and self-optimizing closed-loop collaborative architecture, systematically solving the core defects of idle edge node capabilities, lack of collaboration, and low energy efficiency. Attached Figure Description

[0017] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the accompanying drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. In all the drawings, similar elements or parts are generally identified by similar reference numerals. In the drawings, the elements or parts are not necessarily drawn to scale.

[0018] Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] It should be understood that, when used in this specification and the appended claims, the terms "comprising" and "including" indicate the presence of the described features, integrals, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or collections thereof.

[0021] It should also be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the invention. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.

[0022] It should also be further understood that the term "and / or" as used in this specification and the appended claims refers to any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0023] In one embodiment of the present invention, a method for monitoring and diagnosing potential faults in overhead transmission lines based on edge-end computation is provided, such as... Figure 1 As shown, it includes the following steps: Step 1: Based on environmental characteristics such as altitude and climate, and node distribution density, the monitoring network is dynamically divided into several homogeneous regions with high data cohesion through composite similarity calculation, thus defining the optimal data boundary for subsequent model training.

[0024] The goal of this step is to divide the monitoring network into K optimal regions, ensuring that the environment within each region is homogeneous and the node density is moderate.

[0025] Specifically as follows: S101: First, collect the set of edge node locations. Environmental feature matrix The node communication radius R is defined, and the combined distance between nodes i and j is defined as follows: In the formula, The distance is the geographic Euclidean distance; The Euclidean distance is standardized based on environmental characteristics. Let i be the set of the communicating neighbors of node i. Let j be the set of the communicating neighbors of node j. These are the weighting coefficients, and the sum of the three is 1.

[0026] S102: Perform density clustering partitioning.

[0027] For example, this embodiment uses an improved DBSCAN algorithm, and the implementation steps are as follows: Core point determination: Node i is a core point if and only if its ε-neighborhood contains at least MinPts nodes: Region generation: All reachable core points and their neighboring nodes constitute a region. R k ; Parameter adaptation: e = μd + σd ,in μd The average nearest neighbor distance, σd Standard deviation; This step outputs a set of region partitions. Each region is assigned a unique model .

[0028] Step 2: Train a multimodal fault identification model adapted to the local microenvironment using historical data from each region. Ensure that the model has both generalizability and retains regional specificity through global-local difference constraints, and output model performance indicators for subsequent evaluation.

[0029] The goal of this step is to train a lightweight AI model tailored to local features for each region. Specifically, this includes the following steps: S201: Construct a time-series dataset from the historical data of region k: in, For vibration, leakage current, meteorological, and image multimodal characteristics, For fault labels; S202: Employs a federated transfer learning framework with a local model for region k. Minimize: In the formula, For cross-entropy loss, express, Trade-off coefficient (usually 0.1~0.3); S203: Periodically evaluate the performance of models in each region via the cloud. when Model retraining is triggered when the value is below a set threshold.

[0030] Step 3: Construct a directed weighted graph based on the line impedance matrix and spatial distance, calculate the fault propagation weight of each node to other nodes, and form a quantitative basis matrix for wake-up decision and diagnosis voting.

[0031] The goal of this step is to quantify the electrical influence relationships between nodes, providing a basis for the wake-up strategy. Specifically: S301: Construct a directed weighted graph G = (V, E, W); where vertices V are all edge nodes, edges E represent electrical connections (based on the line impedance matrix Z); and weights W represent the fault propagation coefficient from node i to j. In the formula, This is the attenuation coefficient (typically 0.01~0.05). Let i be the electrical neighbor of node i.

[0032] S302: When node i triggers an early warning, its impact vector is: Set the impact threshold θ (usually 0.2), and the set of nodes to be woken up: Step 4: Calculate the failure rate of each region and introduce the coefficient of variation correction. Combine topology coupling compensation and total power consumption constraints to generate an adaptive model start frequency for each node, so as to achieve optimal energy efficiency scheduling for higher risks and more intensive monitoring.

[0033] The goal of this step is to achieve adaptive monitoring frequency for optimal energy consumption at the regional level. Specifically: S401: Perform regional fault frequency statistics. The fault frequency of region k within the observation period ΔT is used as the representation, as follows: In the formula, This represents the number of fault events in region k. This represents the coefficient of variation for the fault time interval (used to penalize sudden faults). This represents the risk sensitivity coefficient (typically 0.2~0.5).

[0034] S402: Calculate the activation frequency of the fault model, the theoretical sampling frequency of node i in region k: In the formula, This indicates the minimum / maximum sampling frequency set by the system (e.g., 0.1Hz~10Hz). This indicates the electrical influence of node i on its neighbors; This represents the coupling strength coefficient (usually 0.1~0.2).

[0035] S403: Frequency correction based on energy consumption constraints, power consumption model of node i: in, Indicates upload flag, total power consumption constraint: The corrected frequency is determined using the Lagrange multiplier method. Step 5: When a node detects an anomaly, it immediately initiates a three-level wake-up chain of local-regional-cross-regional nodes, activates associated nodes based on the propagation weights in Step 3, and completes the fault determination by merging the inference results of multiple nodes through weighted voting.

[0036] The goal of this step is to achieve fault-triggered chained activation and region-based collaborative reasoning. Specifically: S501: Establish a hierarchical wake-up mechanism, as detailed below: When node i detects an anomaly locally (anomaly score) yes > t (local), then: 4. L1 local wake-up: Node i immediately switches to f max, continuous T burst seconds; 5. L2 Area Wake-up: Broadcasts an alarm to all nodes within area k, affecting the set A of nodes. i according to: 6. L3 cross-region wake-up: If the fault frequency in region k is... Adjacent areas Entering alert mode, frequency increases by 30%.

[0037] S502: Wake-up node set A i Upload features to the edge aggregation node (regional gateway) and perform integrated diagnostics: Among them, weight This reflects the degree of electrical correlation.

[0038] Step Six: Calculate the regional risk score by combining the failure frequency, model aging degree and performance indicators. Based on this, dynamically trigger regional re-division, model retraining, wake-up threshold adjustment and energy consumption quota correction to form a closed-loop self-evolution.

[0039] The goal of this step is to quantify regional risks and optimize resource allocation in reverse.

[0040] S601: Obtain the comprehensive risk score for region k: In the formula, Representation Model Since the last update, This represents the weights, and the sum of the three is 1; S602: According to Adjust the resource quota for region k: For example: High-risk control areas ( >0.7): Increase edge node density This shortens the model update cycle to one week. Medium-risk areas (0.4 < ≤ 0.7): Maintain the current configuration and enable incremental learning; low-risk areas ( ≤ 0.4): Reduce sampling frequency by 20%, and the node goes into daytime sleep mode.

[0041] In summary, this invention overcomes the generalization error bottleneck of traditional homogeneous models by dividing homogeneous regions into homogeneous regions through environment-density clustering and training a dedicated lightweight model. It constructs an electrical topology weighted graph to quantify the fault propagation coefficient between nodes, achieving precise wake-up rather than blind start-up across the entire region, thus reducing ineffective energy consumption. It introduces a fault frequency dynamic scheduling algorithm combined with topology coupling compensation and power budget game theory, enabling the monitoring frequency to adaptively adjust with risk, reducing total system energy consumption while increasing the detection frequency of high-risk areas. Finally, through an event-driven hierarchical wake-up chain and multi-node weighted voting diagnosis, it compresses response latency, forming a self-aware, self-decision-making, and self-optimizing closed-loop collaborative architecture, systematically solving the core defects of idle edge node capabilities, lack of collaboration, and low energy efficiency.

[0042] Those skilled in the art will recognize that the units of the various examples described in connection with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of the invention.

[0043] In the embodiments provided by the present invention, it should be understood that the division of units is only a logical functional division. In actual implementation, there may be other division methods, such as multiple units can be combined into one unit, one unit can be split into multiple units, or some features can be ignored.

[0044] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0045] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.

[0046] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention, and they should all be covered within the scope of the claims and specification of the present invention.

Claims

1. A method for monitoring and diagnosing hidden faults in overhead transmission lines based on edge-end computation, characterized in that, Includes the following steps: S1: Based on environmental characteristics including altitude and climate, and node distribution density, the monitoring network is dynamically divided into several homogeneous regions with high data cohesion through composite similarity calculation, thus defining the optimal data boundary for subsequent model training. S2: Train multimodal fault identification models adapted to the local microenvironment using historical data from each region. Ensure that the model has both generalizability and retains regional specificity through global-local difference constraints, and output model performance indicators for subsequent evaluation. S3: Construct a directed weighted graph based on the line impedance matrix and spatial distance, calculate the fault propagation weight of each node to other nodes, and form a quantitative basis matrix for wake-up decision and diagnosis voting; S4: Calculate the failure rate of each region and introduce the coefficient of variation correction. Combine topology coupling compensation and total power consumption constraints to generate an adaptive model start frequency for each node, so as to achieve optimal energy efficiency scheduling for higher risks and more intensive monitoring. S5: When a node detects an anomaly, it immediately initiates a three-level wake-up chain of local-regional-cross-regional nodes, activates associated nodes based on the propagation weight in step S3, and completes the fault determination by merging the inference results of multiple nodes through weighted voting.

2. The method for monitoring and diagnosing hidden faults in overhead transmission lines based on edge-end calculation as described in claim 1, characterized in that, Step S1 is as follows: S102: Obtain the set of edge node locations P, the environmental feature matrix E, and the node communication radius R, and define the comprehensive distance between nodes i and j, as follows: In the formula, For geographical Euclidean distance; The Euclidean distance is standardized based on environmental characteristics. Let i be the set of the communicating neighbors of node i. Let j be the set of the communicating neighbors of node j. These are the weighting coefficients; S102: Perform density clustering partitioning.

3. The method for monitoring and diagnosing hidden faults in overhead transmission lines based on edge-end calculation as described in claim 1, characterized in that, Step 2 is as follows: S201: Construct a time series dataset from the historical data of region k, and train a multimodal fault identification model adapted to the local microenvironment using the historical data of each region. S202: Employing a federated transfer learning framework, enabling the local model M of region k... k minimize; S203: The cloud periodically evaluates the model performance in each region and decides whether to trigger model retraining based on the evaluation results.

4. The method for monitoring and diagnosing hidden faults in overhead transmission lines based on edge-end calculation as described in claim 1, characterized in that, Step S3 is as follows: S301: Construct a directed weighted graph G = (V, E, W); where vertices V are all edge nodes, edges E represent electrical connections, and weights W represent the fault propagation coefficient from node i to j. In the formula, The attenuation coefficient is... Let i be the electrical neighbor of node i; S302: When node i triggers an early warning, the impact vector is: S303: Set the impact threshold θ, and obtain the set of nodes that need to be woken up based on the threshold: 。 5. The method for monitoring and diagnosing hidden faults in overhead transmission lines based on edge-end calculation as described in claim 1, characterized in that, Step S4 is as follows: S401: Perform area fault frequency monitoring statistics; S402: Calculate the activation frequency of the fault model, the theoretical sampling frequency of node i in region k: In the formula, This indicates the minimum / maximum sampling frequency set by the system; This indicates the electrical influence of node i on its neighbors; Indicates the coupling strength coefficient; S403: Frequency correction based on energy consumption constraints, power consumption model of node i: in, Indicates upload flag, total power consumption constraint: The corrected frequency is determined using the Lagrange multiplier method. 。 6. The method for monitoring and diagnosing hidden faults in overhead transmission lines based on edge-end calculation as described in claim 1, characterized in that, Step S5 is as follows: S501: Establish a hierarchical wake-up mechanism, as detailed below: If node i detects an anomaly locally, then: L1 local wake-up: Node i immediately switches to f max, continuous T burst seconds; L2 Zone Wake-up: Broadcasts an alarm to all nodes within zone k, affecting a set of nodes A. i according to: L3 cross-region wake-up: If the region k fault frequency λk > λ critical, adjacent area Rk +1 Enters alert mode, frequency increases by X%; S502: Wake-up node set A i Upload features to the edge aggregation node and perform integrated diagnostics: in, Indicates the weight.

7. The method for monitoring and diagnosing hidden faults in overhead transmission lines based on edge-end calculation as described in claim 6, characterized in that, Weight Calculated using the following formula: 。 8. The method for monitoring and diagnosing hidden faults in overhead transmission lines based on edge-end calculation as described in claim 1, characterized in that, It also includes step S6: Calculate the regional risk score by combining the failure frequency, model aging degree and performance indicators, and dynamically trigger regional re-division, model retraining, wake-up threshold adjustment and energy consumption quota correction to form a closed-loop self-evolution. Step S6 is as follows: S601: Obtain the comprehensive risk score for region k: In the formula, Representation Model Since the last update, This represents the weights, and the sum of the three is 1; S602: According to Adjust the resource quota for region k.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored program, wherein, when the program is executed, it controls the device containing the computer-readable storage medium to perform the edge-computing-based overhead transmission line fault monitoring and diagnosis method according to any one of claims 1 to 8.

10. A processor, characterized in that, The processor is used to run a program, wherein the program executes the method for monitoring and diagnosing hidden faults in overhead transmission lines based on edge computing as described in any one of claims 1 to 8.

Images