A self-healing control method and system for communication link faults in intelligent substations

Abstract

This invention discloses a method and system for self-healing control of communication link faults in intelligent substations, relating to the field of fault self-healing control technology. The method includes the following steps: constructing a link disturbance intensity sequence reflecting link quality fluctuations; performing trend fitting and abrupt change identification on the link disturbance intensity sequence based on a sliding window mechanism to generate a link status anomaly index; when the link status anomaly index exceeds a first threshold, identifying the affected key vulnerable service set based on the service path topology data of the abnormal link; constructing a context-aware path migration cost model based on the key vulnerable service set and generating an optimal switching path sequence; and implementing hierarchical link control based on the optimal path switching sequence. This invention, by employing an overlapping sliding window mechanism and abrupt change scoring function for trend fitting and abrupt change detection, can effectively identify intermittent or hidden link degradation anomalies that are difficult to detect, thus giving the link status anomaly index higher discrimination accuracy.

Description

Technical Field

[0001] This invention relates to the field of fault self-healing control technology, and more specifically, to a method and system for fault self-healing control of communication links in intelligent substations. Background Technology

[0002] With the development of smart grids, the internal communication network of substations undertakes critical tasks such as equipment status acquisition, protection command transmission, and control coordination. The quality fluctuations and real-time performance of its communication links directly affect the safety and efficiency of power grid operation. Currently, most mainstream communication link status detection methods are based on fixed-period polling mechanisms or preset heartbeat response mechanisms, which can detect serious faults such as link interruptions or complete loss of connection to a certain extent.

[0003] However, in actual operation, communication links more often exhibit weak anomalies such as intermittent fluctuations, minor degradation, or path congestion. These state changes are often not continuous or completely interrupted, making it difficult for traditional methods to identify or respond effectively in a timely manner. On the one hand, fixed polling mechanisms have significant time delays in responding to abnormal changes, failing to meet the high timeliness requirements of critical services. On the other hand, heartbeat-based responses are insensitive to link degradation and have a high false alarm rate for short-term fluctuations, easily leading to false link switching or missed detection of potential faults, resulting in security risks such as delayed self-healing responses or control failures.

[0004] Furthermore, existing self-healing strategies often decouple link anomalies from service impact analysis, lacking accurate impact identification based on service path dependencies, and making it difficult to generate recovery paths with controllable resources and clear control objectives. Therefore, there is an urgent need for a link fault self-healing method that can accurately identify weak link disturbances, dynamically sense abnormal trends, and combine service dependencies for path control, in order to improve the robustness and service continuity assurance capabilities of communication systems.

[0005] The above-disclosed technical solutions have at least the following technical problems: existing methods mostly rely on fixed polling mechanisms or preset heartbeat mechanisms to detect the status of communication links, making it difficult to detect link degradation or weak anomalies in a timely manner; they often cannot effectively distinguish between short-term link fluctuations or intermittent anomalies, which can easily lead to misjudgment or missed judgment, resulting in delayed or ineffective self-healing responses.

[0006] To address the above problems, this invention proposes a solution. Summary of the Invention

[0007] To overcome the aforementioned deficiencies of the prior art, embodiments of the present invention provide a method and system for self-healing control of communication link faults in intelligent substations. By constructing a link disturbance intensity sequence and triggering a critical business path analysis and path control mechanism based on state anomaly indicators, the method addresses the problem that existing methods rely on fixed polling or heartbeat mechanisms, which cannot promptly identify link degradation or weak anomalies, leading to delayed or false triggering of self-healing responses. This enables dynamic perception, accurate identification, and hierarchical control of the communication link status, thereby improving the service continuity and operational robustness of the substation communication system.

[0008] To achieve the above objectives, the present invention provides the following technical solution:

[0009] On the one hand, a self-healing control method for communication link faults in intelligent substations includes the following steps: collecting time-series operational characteristic data and service path topology data of the substation communication links, and constructing a link disturbance intensity sequence reflecting link quality fluctuations; performing trend fitting and abrupt change identification on the link disturbance intensity sequence based on a sliding window mechanism to generate dynamically updated link status anomaly indicators; when the link status anomaly indicators exceed a first threshold, identifying the affected key vulnerable service set based on the service path topology data of the abnormal link; combining the key vulnerable service set, the topological coupling relationship of the abnormal link, and historical control feedback data to construct a context-aware path migration cost model and output the optimal switching path sequence;

[0010] Perform hierarchical link switching operations according to the optimal switching path sequence, collect feedback data on the status of the switched links to evaluate the control results, and dynamically adjust the first threshold based on the evaluation results.

[0011] In a preferred embodiment, the link disturbance intensity sequence reflecting link quality fluctuations comprises the following steps: For each communication link, first data is periodically collected within a preset time window; the first data includes time-series data of transmission delay, packet loss rate, and routing hop count; the collected delay data is subjected to volatility calculation, the packet loss rate time-series data is subjected to instantaneous surge magnitude extraction, and the hop count time-series data is subjected to discontinuity detection, and normalized respectively; the normalized delay volatility, packet loss surge magnitude, and hop count discontinuity index are linearly combined according to preset weight coefficients to generate a disturbance intensity index reflecting the instantaneous instability level of the link; the disturbance intensity indexes at each sampling time are aggregated in chronological order to form a link disturbance intensity sequence.

[0012] In a preferred embodiment, the step of performing trend fitting and mutation identification on the link disturbance intensity sequence based on the sliding window mechanism to generate a dynamically updated link status anomaly index specifically involves: segmenting the link disturbance intensity sequence based on a sliding window mechanism with a preset overlap rate, calculating the mean and variance of the disturbance intensity within each window; fitting the trend of the mean disturbance intensity change of the current window and its predecessor window, and constructing a mutation scoring function by combining the mutation intensity variance of the current window with the mutation rate of historical windows; and generating a link status anomaly index characterizing the abnormal state of the link when the mutation score exceeds a preset scoring threshold.

[0013] In a preferred embodiment, identifying the set of critically vulnerable services affected by abnormal links based on the service path topology data of the abnormal links specifically involves: analyzing the topological position of abnormal link nodes marked by link status anomaly indicators in the service path mapping relationship to generate a link disturbance propagation map characterizing the fault propagation path; for service paths containing abnormal links in the map, calculating the structural dependency parameters of the path relative to the abnormal links based on their path topology position and communication performance parameters; weighting and fusing the structural dependency parameters according to preset weights to generate a path impact strength score; comparing the impact strength scores of each path with a set interference tolerance threshold, filtering the services carried by paths whose scores exceed the threshold, and generating a set of critically vulnerable services.

[0014] In a preferred embodiment, the construction of the context-aware path migration cost model specifically involves: extracting topological information and historical interference co-occurrence data of the network area where the abnormal links are located based on the key susceptible service set, and constructing a link coupling relationship matrix characterizing the fault correlation strength between links; inputting the link coupling relationship matrix and the state anomaly index of the abnormal links into a weighted graph propagation algorithm to generate a structural dynamic propagation chain model, and outputting a high-risk path cluster containing potential cascading fault risks; adjusting the disturbance risk weights based on the high-risk path clusters, and combining the data of each service within the key susceptible service set to construct a context-aware path migration cost model.

[0015] In a preferred embodiment, the generation of the structural dynamic propagation chain model outputs a high-risk path cluster containing potential cascading failure risks. Specifically, this involves: constructing link propagation based on the link coupling relationship matrix and state anomaly indicators; using abnormal link nodes as initial disturbance sources, performing risk propagation through a multi-source random walk algorithm, including defining propagation attenuation factors, path coupling strength coefficients, and path propagation probabilities, calculating the cumulative disturbance risk value of each node, and outputting a structural dynamic propagation chain model characterizing the disturbance diffusion intensity; analyzing the spatial distribution of the cumulative risk value in the propagation chain model, calculating the risk density value within a unit topology region, and filtering the set of paths with risk densities higher than a set threshold to form a high-risk path cluster.

[0016] In a preferred embodiment, the step of adjusting the disturbance risk weight based on high-risk path clusters and constructing a context-aware path migration cost model by combining data from each service within the critically vulnerable service set includes the following steps: marking the disturbance risk level of each link based on the high-risk path cluster, wherein the risk level is determined by statistically analyzing the frequency of each link appearing in the high-risk path cluster; for each candidate path in the path candidate set, accumulating the risk level values ​​of its contained links to generate a path disturbance risk weight value; for each target service in the critically vulnerable service set, filtering reachable paths from the candidate path set and collecting the indicator parameters of each path in real time; and constructing a context-aware path migration cost model based on the indicator parameters.

[0017] In a preferred embodiment, the step of outputting the optimal switching path sequence specifically involves: acquiring historical link switching records and corresponding service recovery feedback data, constructing a heterogeneous graph with path nodes and service recovery quality indicators as elements; learning the correlation between path switching operations and service recovery quality through a graph neural network model, and outputting the expected recovery value of each candidate path for the target service; weightedly fusing the expected recovery value with the output of the path migration cost model, dynamically adjusting the path selection ranking strategy based on the fusion result, and generating the optimal switching path sequence by arranging them in ascending order of migration cost; and dynamically correcting the weight parameters in the path migration cost function based on the recovery effectiveness of different ranking strategies in historical control feedback.

[0018] In a preferred embodiment, the step of performing hierarchical link switching operations according to the optimal switching path sequence, collecting link status feedback data after switching to evaluate the control results, and dynamically adjusting the first threshold based on the evaluation results specifically involves: constructing a preferred path switching sequence; mapping core services to the path with the lowest migration cost based on path migration cost and service level priority, and allocating non-core services to redundant links or rate-limited channels; continuously sampling the link status and service packet transmission status of the switched paths, collecting packet loss rate, retransmission rate, and service recovery latency indicators to form a link control feedback indicator set; constructing a control effectiveness scoring function based on the link control feedback indicator set, and calculating the effectiveness score of the current path switching strategy; comparing the effectiveness score with a preset tolerance threshold, and if the score is lower than the tolerance threshold for multiple consecutive rounds, dynamically adjusting the first threshold for judging link status anomalies based on the scoring trend.

[0019] On the other hand, a smart substation communication link fault self-healing control system includes the following modules: a link disturbance modeling module, used to collect time-series operation characteristic data and service path topology data of the substation communication link, and construct a link disturbance intensity sequence reflecting link quality fluctuations; a state anomaly detection module, used to perform trend fitting and abrupt change identification on the link disturbance intensity sequence based on a sliding window mechanism, and generate dynamically updated link state anomaly indicators; a vulnerable service identification module, used to identify the affected key vulnerable service set based on the service path topology data of the abnormal link when the link state anomaly indicator exceeds a first threshold; a path evaluation module, used to construct a context-aware path migration cost model by combining the key vulnerable service set, the topological coupling relationship of the abnormal link, and historical control feedback data, and output the optimal switching path sequence; and a link control and adaptive feedback module, used to perform hierarchical link switching operations according to the optimal switching path sequence, collect link state feedback data after switching to evaluate the control results, and dynamically adjust the first threshold based on the evaluation results.

[0020] The technical effects and advantages of the self-healing control method and system for communication link faults in intelligent substations according to the present invention are as follows:

[0021] 1. This invention constructs a link disturbance intensity sequence that integrates multiple dimensions such as latency, packet loss rate, and hop count, and uses an overlapping sliding window mechanism and a mutation scoring function for trend fitting and mutation detection. This effectively identifies intermittent or hidden link degradation anomalies that are difficult to capture by traditional polling / heartbeat mechanisms, giving the link status anomaly indicators higher sensitivity and discrimination accuracy. This provides a reliable foundation for subsequent self-healing regulation, achieves highly sensitive perception and accurate identification of weak disturbances in communication links, and improves the timeliness and effectiveness of link anomaly trigger response.

[0022] 2. This invention constructs a migration cost model by integrating contextual information such as link topology coupling relationship, disturbance propagation risk, link resource status and protocol compatibility. It combines graph neural network to predict the impact of path switching on service recovery and performs adaptive weight correction based on historical feedback. It can dynamically generate the optimal switching path sequence driven by service sensitivity. Furthermore, it adjusts the discrimination threshold in reverse through the scoring results to achieve adaptive closed-loop evolution of the control logic, effectively improving the quality fluctuation and adaptability of the self-healing strategy. Attached Figure Description

[0023] Figure 1 This is a flowchart illustrating a self-healing control method for communication link faults in intelligent substations according to the present invention.

[0024] Figure 2 This is a schematic diagram of the structure of a smart substation communication link fault self-healing control system according to the present invention. Detailed Implementation

[0025] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0026] Example 1, Figure 1 This invention discloses a self-healing control method for communication link faults in intelligent substations, comprising the following steps:

[0027] S1, collect time-series operation characteristic data and service path topology data of substation communication links, and construct a link disturbance intensity sequence that reflects link quality fluctuations;

[0028] The time-series operational characteristic data includes delay change sequences, packet loss rate change sequences, hop count change sequences, jitter intensity sequences, and forwarding interval fluctuation sequences.

[0029] The business path topology data includes the forwarding paths of various business messages, the sequence of link nodes involved in the path, the path priority, and the message arrival delay threshold.

[0030] In this embodiment, a link disturbance strength sequence reflecting link quality fluctuations is constructed. The specific steps are as follows:

[0031] For each communication link, time-series data on delay, packet loss rate, and hop count changes are collected within a preset time window, and the sampling results are normalized.

[0032] The normalized latency fluctuation rate, instantaneous packet loss surge magnitude, and hop count discontinuity are linearly combined according to preset weights to obtain a disturbance intensity index that reflects the instantaneous instability level of the link.

[0033] The disturbance intensity indicators corresponding to each time point are arranged in the sampling order to form a link disturbance intensity sequence, which serves as the input basis for subsequent state trend identification and anomaly judgment.

[0034] S2, based on the sliding window mechanism, performs trend fitting and abrupt change identification on the link disturbance intensity sequence to generate dynamically updated link state anomaly indicators, specifically:

[0035] The disturbance intensity sequence is segmented based on an overlapping sliding window, and the mean and variance of the disturbance intensity within each window are calculated respectively.

[0036] By fitting the mean change trend of the perturbation intensity between the current window and its predecessor window, and combining the perturbation intensity variance... To determine the magnitude of the sudden increase, a mutation scoring function is constructed. ,

[0037] When the mutation score exceeds the preset score threshold, a link status anomaly index is generated to trigger path dependency analysis and control response.

[0038] In this embodiment, the mutation scoring function Specifically:

[0039]

[0040] in, This represents the average disturbance intensity of the current sliding window. Let be the average disturbance intensity of the first k steps. The variance of the disturbance intensity within the current window. Let Variance be the variance of the perturbation intensity for the first k steps. To prevent tiny positive numbers with a denominator of 0, , These are preset weighting coefficients used to adjust the contribution weight of the mean and variance to the score.

[0041] In this embodiment, the overlapping sliding window refers to segmenting the disturbance intensity sequence with a step size smaller than the window width, so that there is partial overlap between adjacent windows, thereby realizing continuous observation of the disturbance trend and identification of abrupt changes.

[0042] S3, when the link status anomaly index exceeds the first threshold, identify the set of critical and easily disturbed services affected based on the service path topology data of the abnormal link;

[0043] The first threshold is a reference threshold used to determine whether the link status anomaly indicator triggers service dependency parsing. It can be initially set to a fixed value or obtained through training with historical data. It can be dynamically corrected based on the control feedback to improve the adaptability of anomaly identification sensitivity.

[0044] Furthermore, the critical vulnerable service set refers to a set of services that are most susceptible to abnormal links and whose communication interruptions may pose a significant risk to the stable operation of the system, identified based on the service path dependency structure, service level, and latency tolerance characteristics when the communication links of a substation experience disturbances or degradation trends.

[0045] In this embodiment, identifying the affected key vulnerable service set based on the service path topology data of the abnormal link specifically involves:

[0046] Based on the abnormal link nodes marked in the link status anomaly index, their positions in the business path mapping relationship are extracted, and a link disturbance propagation map is constructed to describe the structural diffusion relationship of link status fluctuations in multiple business paths.

[0047] For each service path containing an abnormal link in the link disturbance propagation map, based on its path topology and communication parameters, calculate the structural dependency parameters of the path relative to the abnormal link. These structural dependency parameters include structural dependency degree, path hop count variation magnitude, proportion of available backup paths, and the maximum tolerable latency of the target service (this item is usually a service configuration item and does not need to be calculated in the scoring; it is considered if the latency is expected after path migration). Less than the preset maximum tolerable latency If the tolerance score is high enough, it is considered tolerable; otherwise, the business will be penalized in the evaluation (fault tolerance score will be lowered).

[0048] The above structural dependency parameters are weighted and fused to form a path impact strength score, which reflects the potential disruptive ability of abnormal links to the service continuity of the path.

[0049] Based on the path impact intensity score and the set interference tolerance threshold, the service set corresponding to the path with a score exceeding the threshold is selected to construct the key easily disturbed service set, which serves as the input basis for the subsequent path switching optimization strategy generation and link resource regulation.

[0050] In this embodiment, the specific formula for calculating the structural dependency is as follows:

[0051]

[0052] Furthermore, the specific formula for calculating the variation in path hop count is as follows:

[0053]

[0054] The specific formula for calculating the proportion of available alternative routes is as follows:

[0055]

[0056] The specific formula for calculating the path impact intensity score is as follows:

[0057]

[0058] in, For structural dependence, This represents the number of alternative paths (i.e., detour paths) in path i that can replace the abnormal link. This represents the variation in the number of hops along the path. This represents the number of hops in the rerouted path. This represents the number of hops in the original path. The proportion of available alternative paths. This represents the number of link segments in the current path that are affected by abnormal links but have switchable links. This represents the total number of path links. For path impact strength scoring, , , , These are the weights of each factor (calculated using the entropy weight method). This is the estimated latency after path migration. The maximum tolerable latency for the target business. This is an indicator function that takes the value 1 if path migration causes latency to exceed the maximum tolerable latency, and 0 otherwise.

[0059] S4 combines the key vulnerable service set, the topological coupling relationship of abnormal links and historical control feedback data to construct a context-aware path migration cost model and output the optimal switching path sequence.

[0060] In this embodiment, a context-aware path migration cost model is constructed as follows:

[0061] Based on the set of key and easily disturbed services, the topology information of the network area where the abnormal link is located and the historical interference co-occurrence data between the links are extracted to construct a link coupling relationship matrix. The link coupling relationship matrix reflects the interference transmission intensity and linkage risk between links.

[0062] Based on the link coupling relationship matrix and combined with the state change index of abnormal links, the weighted graph propagation algorithm is used to deduce the propagation path of disturbances along multiple link paths in the network and their cumulative risks, generate a structural dynamic propagation chain model, and identify high-risk path clusters.

[0063] By adjusting the disturbance risk weights of high-risk path clusters and combining the link idleness, path forwarding latency, and protocol compatibility of each service within the critical and easily disturbed service set, a context-aware path migration cost model is constructed. The disturbance risk weights serve as an important adjustment factor in the cost function, reflecting the risk of path quality fluctuations.

[0064] In this embodiment, the link coupling relationship matrix is ​​constructed as follows:

[0065] Analyze the link-sharing node relationships between the abnormal link node and its neighboring links. Fault coupling frequency and traffic forwarding correlation Calculate the coupling strength between link pairs This forms a coupling relationship matrix that reflects the linkage risks of link status. Based on this matrix, high-risk path segments containing abnormal links are identified.

[0066] The link coupling relationship matrix Coupling strength The calculation formula is:

[0067]

[0068]

[0069]

[0070]

[0071] in For link and links The coupling strength, For link-shared node relationships, The Pearson correlation coefficient is the historical traffic of the two links. For fault coupling frequency, , , These are the weighting coefficients (set based on historical experience), satisfying... , For link The flow rate in the t-th time window for The mean, For link The flow rate in the t-th time window for The mean, This represents the total number of time windows. For link and links The number of windows that fail simultaneously.

[0072] Furthermore, based on the link coupling relationship matrix and combined with the state anomaly indicators of abnormal links, a weighted graph propagation algorithm is used to deduce the propagation path of disturbances along multiple link paths in the network and their cumulative risk, generating a structural dynamic propagation chain model to identify high-risk path clusters, specifically:

[0073] Based on the link coupling relationship matrix and state change indicators, a link propagation graph is constructed. Each link is a node in the diagram. The coupling strength is the edge weight. Abnormal link nodes are assigned additional state anomaly weights as excitation values ​​for disturbance sources. ;

[0074] Based on the multi-source random walk algorithm, abnormal link nodes are used as the initial disturbance source to deduce the path disturbance propagation. During the propagation process, the attenuation factor of each hop is comprehensively considered. Path coupling strength and path propagation probability The cumulative disturbance risk value of each link is output. The cumulative disturbance risk value is the weighted superposition result of the weight of the upper edge of the propagation chain and the path propagation probability, so as to obtain the structural dynamic propagation chain model that reflects the intensity of disturbance diffusion.

[0075] Based on the spatial distribution of cumulative risk values ​​in the propagation chain model, a set of paths with a risk clustering degree higher than a set risk density threshold is extracted to form a high-risk path cluster, which is used to guide subsequent path regulation and avoidance.

[0076] The cumulative disturbance risk value of each link is calculated using the following formula:

[0077]

[0078] in, This represents the cumulative disturbance risk value of the node. As the initial disturbance source The incentive value, This is the initial set of abnormal link nodes. To the source of disturbance To the node The set of all propagation paths, The attenuation factor for propagation, It is the product of path coupling strength. To perturb the propagation probability along the edge (u,v), Let p be the number of hops.

[0079] The step of adjusting the disturbance risk weights through high-risk path clusters, and combining the link idleness, path forwarding latency, and protocol compatibility of each service within the key susceptible service set to construct a context-aware path migration cost model, is as follows:

[0080] Based on the high-risk path cluster, the disturbance risk level of each link is marked, and the risk of each path in the candidate path set is accumulated to obtain the disturbance risk weight value of each candidate path. The disturbance risk level of each link is obtained by counting the number of times each link appears in the high-risk path cluster.

[0081] For each service in the critical and easily disrupted service set, reachable paths are selected from the candidate path set, and the link idle time of each path is collected. Path forwarding delay Protocol compatibility coefficient ;

[0082] The disturbance risk weight, link resource idleness, forwarding latency and protocol compatibility are integrated into a multi-factor path migration cost function to construct a context-aware path migration cost model.

[0083] The path migration cost model is as follows:

[0084]

[0085] in, For path migration cost, , , , These are the preset weighting coefficients for idle time, latency, protocol compatibility, and disturbance risk in path selection (obtained based on the analytic hierarchy process). For link idle time, For path forwarding delay, For protocol adaptation coefficients, This represents the perturbation risk weight value.

[0086] In this embodiment, the optimal switching path sequence is output as follows:

[0087] Acquire and construct a path service status inference graph based on historical link switching and service recovery feedback data, learn the impact of path switching on service recovery quality through graph neural network model, and predict the expected recovery value of each candidate path for the target service.

[0088] The predicted expected business recovery value is weighted and fused with the results of the path migration cost model to dynamically adjust the path optimization and ranking strategy.

[0089] Based on the recovery effectiveness of different path optimization and ranking strategies in historical control feedback, the weight parameters in the path migration cost function are adaptively adjusted, and candidate paths are arranged in ascending order of migration cost to form the optimal switching path sequence.

[0090] In this embodiment, the weight parameters of the path migration cost function are adaptively adjusted based on historical control feedback, specifically including:

[0091] Data on the control effects of various link switching strategies under similar topologies and service types are collected. The service recovery time, path quality fluctuations and interference suppression levels corresponding to each strategy are quantified. A strategy-effect mapping model is constructed, and the weight adjustment of each factor in the path cost function is driven by this model to realize the self-evolution of the path optimization strategy with the system state.

[0092] S5, perform hierarchical link handover operations according to the optimal handover path sequence, collect post-handover link status feedback data to evaluate the control results, and dynamically adjust the first threshold based on the evaluation results, specifically:

[0093] Based on the path switching optimization sequence, services are classified according to the path migration cost. Core services with high criticality and high time sensitivity are prioritized and mapped to the high-quality path with the lowest cost, while non-core services are allocated to redundant links or rate-limited channels to alleviate resource competition.

[0094] The link status and service packet transmission status of the switched path are continuously sampled to obtain the packet loss rate, retransmission rate and service recovery delay of the link after the switch, forming a set of link control feedback indicators.

[0095] A scoring function for the effectiveness of regulation is constructed based on the link regulation feedback indicator set to generate an effectiveness score for the current path switching strategy.

[0096] The performance score is compared with the preset tolerance threshold. If the score is lower than the tolerance threshold for multiple consecutive rounds, the first threshold for judging link status anomalies is dynamically adjusted according to the scoring trend.

[0097] The scoring function for the regulation effectiveness is as follows:

[0098]

[0099] in, To score the effectiveness, Due to the delay in business recovery, For packet loss rate, For retransmission rate, , , These are preset weights (obtained based on historical experience). The maximum allowable delay, , These are the historical baseline values ​​for packet loss rate and retransmission rate, respectively.

[0100] In this embodiment, the performance score is compared with a preset tolerance threshold. If the score is lower than the tolerance threshold for multiple consecutive rounds, the first threshold for judging link status anomalies is dynamically adjusted based on the scoring trend. Specifically:

[0101] The system continuously monitors the scoring results generated after path switching. When the continuous score is lower than the expected scoring threshold, it calculates the scoring deviation trend and dynamically corrects the first threshold for judging link status anomalies based on the trend direction and magnitude. Specifically, when the scoring deviation is continuously negative and exceeds the tolerance threshold, the system lowers the threshold to improve the sensitivity of link anomaly detection; when the scoring deviation is continuously positive, the system raises the threshold to suppress over-regulation, thereby achieving adaptive threshold adjustment.

[0102] Example 2: A smart substation communication link fault self-healing control system, comprising the following modules:

[0103] Link disturbance modeling module: used to collect time-series operation characteristic data and service path topology data of substation communication links, and construct a link disturbance intensity sequence that reflects link quality fluctuations;

[0104] State anomaly detection module: used to perform trend fitting and abrupt change identification on the link disturbance intensity sequence based on the sliding window mechanism, and generate dynamically updated link state anomaly indicators;

[0105] The vulnerable service identification module is used to identify the set of key vulnerable services affected by abnormal links based on the service path topology data of the abnormal links when the link status anomaly index exceeds the first threshold.

[0106] Path evaluation module: It is used to combine the topological coupling relationship of key vulnerable service sets and abnormal links with historical control feedback data to build a context-aware path migration cost model and output the optimal switching path sequence.

[0107] Link control and adaptive feedback module: used to perform hierarchical link switching operations according to the optimal switching path sequence, collect link status feedback data after switching to evaluate the control results, and dynamically adjust the first threshold based on the evaluation results.

[0108] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0109] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, in the form of a computer program product.

[0110] Those skilled in the art will recognize that the modules and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. 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 implementation should not be considered beyond the scope of this application.

[0111] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0112] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0113] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A self-healing control method for communication link faults in intelligent substations, characterized in that, The following steps are involved: Collect time-series operational characteristic data and service path topology data of substation communication links, and construct a link disturbance intensity sequence that reflects link quality fluctuations; Based on the sliding window mechanism, trend fitting and abrupt change identification are performed on the link disturbance intensity sequence to generate dynamically updated link state anomaly indicators. When the link status anomaly index exceeds the first threshold, the key vulnerable service set affected by the abnormal link is identified based on the service path topology data of the abnormal link. By combining the key vulnerable service set, the topological coupling relationship of abnormal links, and historical control feedback data, the topological structure information and historical interference co-occurrence data of the network area where the abnormal links are located are extracted, and a link coupling relationship matrix characterizing the fault correlation strength between links is constructed. Based on the link coupling relationship matrix and the state anomaly index of abnormal links, a weighted graph propagation algorithm is input, with abnormal link nodes as the initial disturbance source. Risk propagation is carried out through a multi-source random walk algorithm, including defining the propagation attenuation factor, path coupling strength coefficient, and path propagation probability. The cumulative disturbance risk value of each node is calculated to generate a structural dynamic propagation chain model. The spatial distribution of the cumulative risk value in the propagation chain model is analyzed, and the risk density value in a unit topology area is calculated to screen the set of paths with risk density higher than a set threshold, forming a high-risk path cluster. Based on the adjustment of disturbance risk weights for high-risk path clusters, and combined with data from each business within the set of critical and easily disturbed businesses, a context-aware path migration cost model is constructed to output the optimal switching path sequence. Perform hierarchical link switching operations according to the optimal switching path sequence, collect feedback data on the status of the switched links to evaluate the control results, and dynamically adjust the first threshold based on the evaluation results.

2. The intelligent substation communication link fault self-healing control method according to claim 1, characterized in that, The specific steps for obtaining the link disturbance strength sequence reflecting link quality fluctuations are as follows: For each communication link, first data is periodically collected within a preset time window; the first data includes time-series data of transmission delay, packet loss rate, and routing hop count. The volatility of the collected delay data is calculated, the instantaneous surge magnitude of the packet loss rate time series data is extracted, the discontinuity of the hop count time series data is detected, and normalization is performed on each of them. The normalized latency volatility, packet loss surge magnitude, and hop count discontinuity index are linearly combined according to preset weighting coefficients to generate a disturbance intensity index that reflects the instantaneous instability level of the link. The disturbance intensity indices at each sampling time are aggregated in chronological order to form a link disturbance intensity sequence.

3. The self-healing control method for communication link faults in intelligent substations according to claim 2, characterized in that, The process of performing trend fitting and abrupt change identification on the link disturbance intensity sequence based on the sliding window mechanism to generate dynamically updated link state anomaly indicators is as follows: Based on a sliding window mechanism with a preset overlap rate, the link disturbance intensity sequence is segmented, and the mean and variance of the disturbance intensity within each window are calculated respectively. Fit the mean change trend of the perturbation intensity of the current window and its predecessor windows, and combine the variance of the perturbation intensity of the current window with the magnitude of the sudden increase of the historical makers to construct a mutation scoring function; When the mutation score exceeds the preset score threshold, a link status anomaly index is generated to characterize the abnormal state of the link.

4. The self-healing control method for communication link faults in intelligent substations according to claim 3, characterized in that, The step of identifying the affected key vulnerable service set based on the service path topology data of the abnormal links specifically includes: Based on the abnormal link nodes marked by the link status anomaly index, analyze their topological position in the business path mapping relationship and generate a link disturbance propagation map that represents the fault propagation path. For a service path containing abnormal links in the graph, the structural dependency parameters of the path relative to the abnormal links are calculated based on its path topology location and communication performance parameters. The structural dependency parameters are weighted and fused according to preset weights to generate a path impact intensity score; By comparing the impact intensity scores of each path with the set interference tolerance threshold, the services carried by the paths whose scores exceed the threshold are selected, and a set of critical and easily disturbed services is generated.

5. The intelligent substation communication link fault self-healing control method according to claim 4, characterized in that, The specific steps for adjusting the disturbance risk weights based on high-risk path clusters and constructing a context-aware path migration cost model by combining data from each service within the key susceptible service set are as follows: The disturbance risk level of each link is marked based on the high-risk path cluster, and the risk level is determined by statistically analyzing the frequency of each link in the high-risk path cluster. For each candidate path in the path candidate set, the risk level values ​​of its contained links are accumulated to generate a path disturbance risk weight value; For each target business in the key and easily disrupted business set, reachable paths are selected from the candidate path set, and the indicator parameters of each path are collected in real time. A context-aware path migration cost model is constructed based on indicator parameters.

6. The intelligent substation communication link fault self-healing control method according to claim 5, characterized in that, The output optimal switching path sequence is specifically as follows: Obtain historical link switching records and corresponding service recovery feedback data, and construct a heterogeneous graph with path nodes and service recovery quality indicators as elements; The graph neural network model is used to learn the relationship between path switching operations and service recovery quality, and output the expected recovery value of each candidate path for the target service. The expected recovery value and the output of the path migration cost model are weighted and fused. Based on the fusion result, the path selection and ranking strategy is dynamically adjusted, and the optimal switching path sequence is generated by arranging the paths in ascending order of migration cost. Based on the recovery effectiveness of different sorting strategies in historical regulation feedback, the weight parameters in the path migration cost function are dynamically adjusted.

7. The intelligent substation communication link fault self-healing control method according to claim 6, characterized in that, The step involves performing hierarchical link handover operations based on the optimal handover path sequence, collecting post-handover link status feedback data to evaluate the control results, and dynamically adjusting the first threshold based on the evaluation results. Specifically: Construct a path switching optimization sequence, and map core businesses to the path with the lowest migration cost based on path migration cost and business level priority, while allocating non-core businesses to redundant links or rate-limited channels; The link status and service packet transmission status of the switched paths are continuously sampled to collect packet loss rate, retransmission rate and service recovery delay indicators, forming a link control feedback indicator set; A scoring function for the effectiveness of regulation is constructed based on the link regulation feedback indicator set, and the effectiveness score of the current path switching strategy is calculated. The performance score is compared with the preset tolerance threshold. If the score is lower than the tolerance threshold for multiple consecutive rounds, the first threshold for judging link status anomalies is dynamically adjusted according to the scoring trend.

8. A system using the intelligent substation communication link fault self-healing control method as described in any one of claims 1-7, characterized in that, Includes the following modules: Link disturbance modeling module: used to collect time-series operation characteristic data and service path topology data of substation communication links, and construct a link disturbance intensity sequence that reflects link quality fluctuations; State anomaly detection module: used to perform trend fitting and abrupt change identification on the link disturbance intensity sequence based on the sliding window mechanism, and generate dynamically updated link state anomaly indicators; The vulnerable service identification module is used to identify the set of key vulnerable services affected by abnormal links based on the service path topology data of the abnormal links when the link status anomaly index exceeds the first threshold. Path evaluation module: It is used to combine the topological coupling relationship of key vulnerable service sets and abnormal links with historical control feedback data to build a context-aware path migration cost model and output the optimal switching path sequence. Link control and adaptive feedback module: used to perform hierarchical link switching operations according to the optimal switching path sequence, collect link status feedback data after switching to evaluate the control results, and dynamically adjust the first threshold based on the evaluation results.

Images