A digital base station establishment method for railway operation and maintenance

Abstract

This invention relates to the field of railway operation and maintenance technology, specifically a method for establishing a digital infrastructure for railway operation and maintenance. The method includes: collecting railway operation and maintenance logs and constructing an operation and maintenance topology map composed of multiple grid nodes; using the operating status of each grid node in the topology map as a guide to determine the connection path of each grid node when an alarm event occurs, and determining the association rules corresponding to each connection path; based on the association rules of each connection path, deriving the range boundary corresponding to the alarm event, and dividing the topology map into multiple target subgraphs based on the range boundary; based on the multidimensional data of the target subgraphs, clarifying the execution strategy of each target subgraph, and mapping the response time and status update time of each grid node in the target subgraph when executing the strategy to the alarm handling strategy of each grid node; generating a handling process in the railway operation and maintenance scenario according to the development trajectory of the alarm event using the alarm handling strategy; thus improving railway maintenance efficiency and processing accuracy.

Description

Technical Field

[0001] This invention relates to the field of railway operation and maintenance technology, specifically a method for establishing a digital base for railway operation and maintenance. Background Technology

[0002] With the large-scale expansion of railway networks and the widespread application of intelligent equipment, railway operation and maintenance scenarios exhibit core characteristics such as diverse equipment types, wide node distribution, massive data volume, and complex fault correlations. On the one hand, operation and maintenance data covers multi-source heterogeneous data such as equipment operation logs, maintenance operation logs, approval process logs, and alarm logs, with data dimensions scattered and lacking effective integration. On the other hand, fault events often exhibit characteristics of multi-node correlation, cross-regional propagation, and hidden root causes, leading to a significant decrease in railway operation and maintenance efficiency when alarm events occur.

[0003] For example, Chinese Patent Publication No. CN118689693A discloses a method, apparatus, device, and storage medium for analyzing operation and maintenance tasks. The method includes: acquiring the status information of the operation and maintenance tasks in operation; filtering out target operation and maintenance tasks whose current operating status is abnormal based on the status information and acquiring log information related to the target operation and maintenance tasks; matching the log information related to the target operation and maintenance tasks with the log set corresponding to the feedback content in the preset rule base to obtain a matching result; wherein the feedback content includes cause analysis content and processing suggestion content; if the matching result indicates a successful match, sending the target feedback content in the matching result to the target user so that the target user can analyze the target operation and maintenance task based on the target feedback content; if the matching result indicates a failed match, generating the feedback content corresponding to the target operation and maintenance task and storing it in the rule base.

[0004] For example, Chinese Patent Publication No. CN119782100A discloses an intelligent generation method for operation and maintenance scripts based on a large model, which relates to the field of network operation and maintenance technology. It includes the process of building an asset database and establishing an operation and maintenance network topology diagram, performing word segmentation, extraction and matching based on operation and maintenance requirements actively input by manual operation and maintenance or recommended by automatic operation and maintenance analysis, building an indicator-parameter form, and selecting script modules according to the operating environment of the target asset, filling them according to the indicator-parameter form, and deploying the script to the target asset.

[0005] In existing technologies, the feedback target of operation and maintenance tasks is determined by matching the coding of logs; and the script deployment in the operating environment is determined by the correlation between unfamiliar alarms and other alarms. These processing methods are equivalent to matching processing for log feedback. If railway operation and maintenance occurs, in addition to matching processing based on log feedback, it is also necessary to further filter the location of the alarm and its corresponding association in order to reduce the impact of the fault on railway operation and ultimately achieve the goal of fault migration and self-healing. Summary of the Invention

[0006] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is: a method for establishing a digital base for railway operation and maintenance, including: S1, collecting railway operation and maintenance logs, and constructing an operation and maintenance topology map composed of multiple grid nodes based on the railway line.

[0007] S2 uses the running status of each grid node in the operation and maintenance topology diagram as a guide to determine the connection path of each grid node when an alarm event occurs, estimates the association rules for each connection path, and establishes the association rules corresponding to each connection path.

[0008] S3, based on the association rules of each connection path, derives the range boundary corresponding to the alarm event, and divides the operation and maintenance topology map into multiple target subgraphs based on the range boundary.

[0009] S4. Based on the multidimensional data of the target subgraph, the execution strategy of each target subgraph is defined. The response time and status update time of each grid node in the target subgraph when executing the strategy are mapped to the alarm handling strategy of each grid node.

[0010] S5 generates a handling process for railway operation and maintenance scenarios based on the development trajectory of alarm events, according to the alarm handling strategy.

[0011] The beneficial effects of this invention are as follows: First, this invention collects operation and maintenance logs based on railway lines and maps them to grid nodes. Combining the dual time dimensions of alarm detection time and alarm confirmation time, and through preset logical rules such as time sequence proximity judgment and duration anomaly analysis, it establishes causal relationships between grid nodes and constructs an operation and maintenance topology map; it clarifies the delayed alarms and other situations that occur in the corresponding equipment when an alarm occurs, providing a precise basic data carrier for subsequent correlation analysis and scope definition.

[0012] Second, this invention establishes three types of association rules for nodes on the connection path based on the running status of grid nodes: for abnormal nodes without direct connection but with similar time sequences, it outputs state coexistence rules; for abnormal nodes with upstream and downstream topological connections, it outputs state propagation rules through device attribute query and priority sorting; for upstream nodes where multiple downstream nodes converge, it determines the root cause based on time sequence association and outputs root cause coexistence rules. This enables current operation and maintenance personnel to quickly locate abnormal association patterns and provides logical support for inferring the scope of alarm impact.

[0013] Third, this invention derives the alarm impact range boundary based on association rules. For three scenarios—complete overlap, partial overlap, and no overlap—it uses direct output, priority unified boundary, and regional diagnosis methods to divide the operation and maintenance topology map into multiple target sub-graphs. This further clarifies the device distribution in the operation and maintenance scenario, enabling operation and maintenance resources to be monitored in different areas, thereby improving resource utilization efficiency.

[0014] Fourth, this invention clarifies the executable strategies of the subgraph by using multi-dimensional data based on the target subgraph, such as node attributes, alarm characteristics, and resource status. It uses response time + status update time as the event correlation index and maps it to the alarm handling strategy of each grid node, while synchronizing the correlation data of node running status changes. This allows the strategy to be directly matched with the alarm event scenario, enabling operation and maintenance personnel to quickly initiate handling actions according to standardized strategies. Attached Figure Description

[0015] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0016] Figure 1 This is a flowchart illustrating a method for establishing a digital base station for railway operation and maintenance.

[0017] Figure 2 This is a flowchart illustrating step S1 of a digital base station establishment method for railway operation and maintenance.

[0018] Figure 3 This is a flowchart illustrating step S2 of a digital base station establishment method for railway operation and maintenance.

[0019] Figure 4 This is a flowchart illustrating step S3 of a digital base station establishment method for railway operation and maintenance. Detailed Implementation

[0020] The embodiments of the present invention are described in detail below. The embodiments described below are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention. Where specific techniques or conditions are not specified in the embodiments, they shall be performed in accordance with the techniques or conditions described in the literature in the art or in accordance with the product manual.

[0021] See Figure 1 A method for establishing a digital base for railway operation and maintenance includes: S1, collecting railway operation and maintenance logs and constructing an operation and maintenance topology map composed of multiple grid nodes based on railway lines.

[0022] S2 uses the running status of each grid node in the operation and maintenance topology diagram as a guide to determine the connection path of each grid node when an alarm event occurs, estimates the association rules for each connection path, and establishes the association rules corresponding to each connection path.

[0023] S3, based on the association rules of each connection path, derives the range boundary corresponding to the alarm event, and divides the operation and maintenance topology map into multiple target subgraphs based on the range boundary.

[0024] S4. Based on the multidimensional data of the target subgraph, the execution strategy of each target subgraph is defined. The response time and status update time of each grid node in the target subgraph when executing the strategy are mapped to the alarm handling strategy of each grid node.

[0025] S5 generates a handling process for railway operation and maintenance scenarios based on the development trajectory of alarm events, according to the alarm handling strategy.

[0026] The collected railway operation and maintenance logs will be processed primarily for fault operation and maintenance scheduling. Data such as track sections, time, personnel, and maintenance content will be extracted from the railway operation and maintenance logs, including equipment operation logs, approval process logs, and alarm logs. The operating parameters of signal equipment, track equipment, and power supply equipment on the corresponding track sections will be identified to complete the operation and maintenance monitoring of each track section.

[0027] Equipment operation logs are collected from embedded monitoring systems of railway signaling equipment, track equipment, and power supply equipment, such as signal monitoring terminals, track circuit detectors, and power supply systems; approval process logs are collected from railway construction management systems and approval workflow platforms, such as records of construction plan submissions, approval form issuances, and railway bureau instructions; alarm logs are collected from the alarm platform of the railway operation and maintenance monitoring center, such as information on hardware equipment alarms, database alarms, and application system alarms.

[0028] The core objective of step S1 is to complete the data collection and structured processing of railway operation and maintenance logs, and to build a visualized and traceable operation and maintenance topology map based on the railway line, providing data and spatial topology support for subsequent fault operation and maintenance scheduling.

[0029] like Figure 2 As shown, the implementation of step S1 includes: S11, in response to railway operation and maintenance needs, querying related fields from the railway operation and maintenance logs using preset query fields, and determining them as candidate query fields.

[0030] S12, based on the content corresponding to the candidate query fields, map each candidate query field according to the railway line to form multiple grid nodes.

[0031] S13: Associate each grid node with its topological connection relationship to construct an operation and maintenance topology diagram for multiple grid nodes.

[0032] First, based on the description of the current railway operation and maintenance needs, the query fields are identified from the railway operation and maintenance needs. These query fields are generally preset words, or fields selected according to the current needs, such as train number, line name, equipment ID, location, timestamp, event type, maintenance status, and other preset query fields. After the query fields are entered, all preset query fields and their corresponding values ​​will be found from the railway operation and maintenance logs and set as candidate query fields.

[0033] Then, the corresponding locations in the candidate query fields are associated with specific geographic spaces to form grid nodes with location information; for example, trains are mapped by the line name and kilometer markers, equipment is mapped by the line name and station / kilometer markers, and events are mapped by the associated entities, such as the location of trains or equipment, to form multiple nodes containing grids.

[0034] Finally, the topological connection relationship is to find the inherent connections between these grid nodes, such as geographical proximity, or the existence of causal and hierarchical relationships, and then associate these grid nodes.

[0035] Geographical proximity means that the line where its grid node is located can be directly connected to other grid nodes. In this case, a 1km geographical unit can be set up, or a kilometer marker can be selected to divide a geographical unit into geographical units to set up the grid of the operation and maintenance topology map.

[0036] Causal association means that one event is caused by another event. For example, if the train delay is caused by a signal failure ahead, then there is a causal relationship between two grid nodes. The corresponding grid nodes are then connected and synchronized to the operation and maintenance topology.

[0037] Membership association means that the grid nodes corresponding to trains or equipment belong to a larger line entity. For example, if train nodes and equipment nodes both belong to the larger line entity of Beijing-Shanghai High-Speed ​​Railway, the corresponding grid nodes can be synchronized to the line entity according to their train or equipment form to complete the operation and maintenance topology under membership association.

[0038] Each output grid contains grid code, road segment information, a list of associated devices (including device model / parameters / location), and an operation and maintenance log index. Based on the association of these data, an operation and maintenance topology map is created for the current operation and maintenance requirements.

[0039] Meanwhile, in step S13, when identifying the topological connection relationship, it is also necessary to obtain the tags of each road segment during maintenance, such as alarm event tags, device tags, and neighbor association tags, and record the topological structure related to these tags in order to facilitate data backtracking.

[0040] Therefore, the implementation of step S13 also includes: S131, when receiving railway operation and maintenance requirements, determining the alarm detection time and alarm confirmation time of the grid node.

[0041] S132, based on the alarm detection time and alarm confirmation time of each grid node, perform causal relationship judgment on the grid nodes according to preset logic rules.

[0042] S133: Connect the grid nodes that have completed the judgment and synchronize them to the operation and maintenance topology diagram to complete the causal relationship of multiple grid nodes under the alarm event. The causal relationship data will be synchronized to the node attributes of the operation and maintenance topology diagram. Step S2 can determine the connection path by reading the causal relationship identifier in the node attributes.

[0043] Alarm detection time represents the time from the actual occurrence of a fault to the system automatically detecting and generating an alarm event. Alarm confirmation time represents the time from alarm generation to the dispatcher or maintenance personnel confirming the validity of the alarm, thereby determining the topological connection relationship between each grid node under fault scheduling.

[0044] The implementation of step S132 also includes: for any two grid nodes, a preliminary judgment is made based on the timestamps corresponding to the alarm detection time and alarm confirmation time to determine whether the alarm events between the corresponding grid nodes are temporally adjacent; regardless of whether they are temporally adjacent, a secondary judgment process will be entered, and the judgment will be made sequentially according to the durations corresponding to the alarm detection time and alarm confirmation time.

[0045] For the grid nodes after the initial judgment, a second judgment is made based on the duration corresponding to the alarm detection time and alarm confirmation time. Based on the judgment results, the causal relationship between the corresponding grid nodes is set.

[0046] When setting up causal relationships, the grid nodes are not directly connected. Instead, they are based on preset logical rules that utilize the duration and timing information of alarm detection time and alarm confirmation time.

[0047] Rule 1, Temporal Proximity Association: If the alarm confirmation time of node a is very close to the alarm detection time of node b, such as within 5 or 15 minutes, and the two are geographically adjacent, then it is considered that the fault of a may be the cause of the fault of b, or that both belong to a larger range of faults. In this case, a directed dashed line is drawn from a to b on the topology graph, labeled as temporal proximity, and the time difference is marked, such as b being generated 3 minutes after a is confirmed. The time threshold for temporal proximity association is based on the historical alarm data statistics of the corresponding railway line. For example, the default threshold for high-speed railway lines is 5 minutes, and the default threshold for conventional railways is 15 minutes. It can be dynamically adjusted according to the line maintenance level and the average value of historical data statistics.

[0048] Rule 2. Correlation of Abnormal Detection Duration: If the alarm detection duration of a certain node 'a' is much higher than its historical average or the average of similar nodes, for example, normally 2 minutes, but 'a' is 10 minutes, and its downstream node 'b' has a related alarm in a short period of time, then the detection delay of 'a' may have caused the failure of 'b' to be prevented in time. In this case, a directed solid line is established from 'a' to 'b', labeled as the impact of detection delay, and the alarm detection data of 'a' is attached.

[0049] Rule 3: Correlation of Abnormal Confirmation Time: If the confirmation time for alarms from multiple nodes within a certain area is generally long, it indicates that there may be a bottleneck in the scheduling response of that area. A high-latency correlation cluster can be established among these nodes; these nodes are enclosed with boxes of the same color to form a latency cluster, and the average confirmation time is marked as 8 minutes, to complete the data synchronization of causal correlation to the operation and maintenance topology diagram.

[0050] These rules are confirmed by the timestamp of the alarm event. Using time and geographical location as dimensions, the relevant devices of the current alarm event are initially aggregated, and the relevant grid nodes are marked on the operation and maintenance topology map to facilitate subsequent adjustment and processing of grid nodes according to the operating status.

[0051] It should be noted that the preset logical rules for temporal proximity association, detection duration abnormal association, and confirmation duration abnormal association will call the preset time values ​​and mark the time values ​​at which the corresponding abnormalities occur in different positions.

[0052] Furthermore, in the initial judgment, the alarm confirmation time of any grid node and the alarm detection time of another node are checked to see if their timestamps are close. Based on the average time interval between alarm confirmation time and alarm detection time at adjacent geographical locations in historical data, adjacent grid nodes with intervals smaller than this interval are considered as temporally close associations as described at this time; if they are larger than this interval, a secondary judgment is made on the subsequent detection duration and confirmation duration.

[0053] During the secondary judgment, the average time of alarm confirmation and alarm detection over multiple detections is used as the threshold. If it exceeds this value, it is considered that there is an abnormal correlation. Furthermore, if these nodes exist in multiple locations in a certain area, the corresponding data is marked with boxes. Finally, the abnormal correlation identification of each grid node under the condition of temporal proximity is completed.

[0054] In one embodiment of the present invention, in step S2, the path of the operation and maintenance topology graph is abstracted, that is, a sequence of a series of nodes and edges in the graph. Data such as a complete train operation process and a fault propagation chain are analyzed and identified to complete the logical evolution of the corresponding data.

[0055] The operational status in the maintenance topology diagram will be described using a binary classification of abnormal and normal, explaining the status of the equipment mapped to the corresponding grid nodes under maintenance and operation. The identified association rules will be explained according to three dimensions: state co-occurrence, state propagation, and root cause co-occurrence, to explain the association rules in the railway maintenance process.

[0056] State coexistence occurs when multiple non-directly connected nodes frequently enter abnormal states simultaneously or successively. In this case, the corresponding grid nodes may share a power supply, transmission equipment, or service system, and it is necessary to determine the correlation between the abnormal alarms of these non-directly connected nodes.

[0057] State propagation means that if the state of node a changes from normal to abnormal, and within a very short time, the state of its downstream node b also changes from normal to abnormal, then there is a state propagation path from a to b, which directly reflects the transmission path of the fault.

[0058] Root cause symbiosis means that if multiple downstream paths that converge to the same upstream node report abnormal states at similar times, then their common upstream node is very likely the root cause of the failure.

[0059] These three dimensions—time proximity and co-occurrence—are used in data mining to determine the association rules contained in the current operation and maintenance topology.

[0060] Meanwhile, alarm events can be categorized by source into national railway alarms and railway bureau alarms; and by object into hardware equipment alarms, database alarms, middleware alarms, application alarms, terminal and other alarms, and load balancing alarms. Each type of alarm is assigned a unique task field (task name, handling level, assigned position, initiator, initiation time, current stage, handler, task source, first time, alarm level, alarm grid) to determine the specific circumstances of alarm events occurring under railway operation and maintenance.

[0061] like Figure 3 As shown, the implementation of step S2 includes: S21, for multiple grid nodes on the connection path, selecting grid nodes with the same business scenario to generate a grid node set. Business scenarios are divided according to the combination of line type (high-speed rail or conventional rail, etc.) + equipment function (signal / track / power supply), and the same business scenario means that both the line type and equipment function are the same.

[0062] S22: For abnormal nodes in the set of grid nodes that have no direct connection relationship, determine whether the abnormal node satisfies the temporal proximity association. If it does, output the corresponding grid node as a state co-occurrence rule.

[0063] S23. For the remaining nodes in the set of grid nodes, obtain the upstream and downstream nodes of each grid node in the connection path. Based on the topological relationship of the upstream and downstream nodes, filter the grid nodes that meet the anomaly propagation conditions and output them as state propagation rules.

[0064] S24. When an upstream node corresponds to multiple abnormal downstream nodes, determine whether there is a temporal proximity association between the downstream nodes. If so, determine that the upstream node is the root cause. Based on the running status of the upstream node, aggregate and combine the corresponding downstream nodes and output the root cause co-occurrence rule.

[0065] S25. According to the grid nodes corresponding to the state co-occurrence rule, state propagation rule, and root cause co-occurrence rule, the corresponding data is stored as the association rule corresponding to each connection path.

[0066] During the current connection step, the operating status of each grid node is marked (normal / abnormal), the operating status of all grid nodes is summarized, and the abnormal nodes when the alarm event occurs are recorded. Based on the node sequence of the physical link of the device, the node sequence of the train operation or maintenance business, and the node sequence that may cause the fault to spread according to the abnormal node (e.g., if device a has had a problem, then the connected devices b and c are pre-connected), these node sequences are used as the connection path at this time to illustrate the connection path under the physical link and business scenario.

[0067] When filtering grid nodes that meet the abnormal propagation conditions in step S23, the implementation method includes: extracting the device attributes corresponding to each grid node in the connection path, performing topology relationship query based on the device attributes, and determining the upstream and downstream nodes of the current grid node; the device attributes represent the device type, rated parameters, runtime, historical fault frequency, and maintenance cycle extracted from the device operation log.

[0068] The priorities of upstream and downstream nodes are determined based on their execution order within the business scenario. The priorities between upstream and downstream nodes are determined by examining the execution order of their corresponding devices within the business scenario.

[0069] The highest priority grid node is selected as the node to be verified. If the node to be verified is abnormal, all reachable grid nodes are marked as candidate propagation nodes, starting from the node to be verified.

[0070] The abnormal propagation probability of candidate propagation nodes is determined by the abnormal frequency of candidate propagation nodes, and each connection path is sorted according to the abnormal propagation probability; the currently recorded abnormal propagation probability is the input data used for subsequent state propagation calculation.

[0071] Repeatedly retrieve the highest priority grid node until all grid nodes have been checked.

[0072] The marked candidate propagation nodes and the abnormal nodes to be checked are considered as outputs of the state propagation rules.

[0073] When setting the current association rules, grid nodes with abnormal propagation will be identified, and the propagation probability of abnormalities on these nodes will be recorded. The abnormal propagation probability represents the ratio of the abnormal frequency of candidate propagation nodes to the average abnormal frequency of nodes of the same type. The statistical period for abnormal frequency is 24 hours. Nodes of the same type are divided into three categories according to equipment function: signal equipment, track equipment, and power supply equipment, which facilitates subsequent verification of the possible impact range of alarm events. When using candidate propagation nodes to sort the corresponding connection paths, the relative situation of abnormality in each connection path is quantified by the ratio of the sum of the abnormal propagation probabilities identified on each connection path to the total abnormal propagation probability.

[0074] In one embodiment of the present invention, when configuring the target subgraph, the target subgraph will be selected based on the current operation and maintenance requirements. When the current operation and maintenance requirements are based on the fault operation and maintenance scheduling dimension when the alarm event occurs, the subsequent scheduling processing method will be determined according to the impact range of the alarm event.

[0075] For example, when association rules can be represented by three dimensions: state coexistence, state propagation, and root cause coexistence, the scope of influence identified will be set according to the association rules corresponding to each grid node, and the scope boundary corresponding to the scope of influence will be set. In the association rules of state coexistence, shared resources under the business scenario will be extracted and set, such as shared distribution box A, shared fiber optic link B, shared scheduling system C, etc. These shared resources are used to anchor devices without direct topological connections, and the scope boundary of state coexistence will be quantified according to the physical or logical coverage of the shared resources, such as distribution box A covering grids N200-N206 and fiber optic link B covering grids N201-N208.

[0076] The range boundary of state propagation is then described by the optimal propagation path, such as the path N100 (power supply) → N101 (signal) → N103 (track). The path with the highest score derived from the anomaly propagation probability is considered the range boundary of state propagation.

[0077] When performing state propagation calculation, the conditional probability of two adjacent grid nodes during path advancement is calculated based on the anomaly propagation probability corresponding to each grid node. For example, the conditional probability P(a|b) of grid nodes a and b is calculated. All grid nodes on the current connection path under the same state propagation rule are calculated. The connection path with the largest sum of these conditional probabilities is regarded as the optimal state propagation path, representing the path with the highest probability of anomaly propagation under the current alarm event. The grid nodes corresponding to this path are regarded as the output range boundary, further defining the association between each device.

[0078] Root cause symbiosis relies on its upstream node, and gathers all downstream nodes and child nodes of the upstream node as its range boundary. At this time, multiple range boundaries are obtained by relying on association rules, and the repeated parts of the range boundaries are defined to divide the target subgraph covering multiple grid nodes. Each target subgraph represents the node sequence associated with the alarm event.

[0079] like Figure 4 As shown, the implementation of step S3 includes: S31, setting the scope boundaries of each association rule based on the association rules of each connection path.

[0080] S32, if the boundaries of each association rule completely overlap, output the grid nodes corresponding to each association rule directly as the target subgraph for partitioning.

[0081] S33. If the boundaries of each association rule partially overlap, the intersection of the boundaries is used to determine the priority. The boundary with the highest priority is used to unify the boundaries of each association rule and output as the target subgraph.

[0082] When determining priorities, the priority can be ranked as follows: State propagation priority > Root cause symbiosis priority > State symbiosis priority. Priority setting is based on the scope of fault impact and the urgency of handling. State propagation directly reflects the fault propagation path, with the widest impact and highest urgency; root cause symbiosis focuses on the root cause of the fault, with the next highest urgency; State symbiosis is only an indirect association, with the lowest urgency. Then, the scope of state propagation is used as the primary identification scope. Grid nodes that depend on its upstream nodes and grid nodes without direct topological connections are used as supplementary data under the current state propagation, and these data are unified into the output target subgraph to interpret the size of the area mainly affected by the alarm event.

[0083] If one or more grid nodes belong to the influence area of ​​multiple association rules, they are uniformly described according to their corresponding association rules. Then, it is verified whether the grid nodes corresponding to each association rule are in the area that the current alarm event can cover, such as the continuous area that the maintenance operation can cover, to avoid scattered and isolated nodes. After removing nodes that are irrelevant to the current alarm event and scattered and isolated nodes, the scope boundaries of each association rule are unified, and its data is output as the target subgraph.

[0084] S34. If the boundaries of each association rule do not overlap, determine the number of grid nodes within each boundary, perform regional diagnosis on each boundary based on the number of grid nodes, and regard the diagnosed data as the output target subgraph.

[0085] Generally, each grid node will perform judgment and processing of three association rules: state coexistence, state propagation, and root cause coexistence. When the scope of influence does not overlap, it means that the fault is physically isolated or actively migrated, so that the alarm event that appears is safely isolated and migrated, and the whole system can achieve fault self-healing migration.

[0086] The implementation of the area diagnosis in step S34 also includes: for any grid node within a range boundary, determining the number of grid nodes within the corresponding range boundary; and when the number of grid nodes has not changed within at least three consecutive time periods, marking the corresponding location as an isolation segment using the alarm event corresponding to the current association rule.

[0087] The probability of anomaly propagation for each grid node within the isolation zone is statistically analyzed. When the probability of anomaly propagation for each grid node becomes stable, the area containing that grid node is considered the target subgraph in the output. The isolation zone indicates that policy control has been implemented in the corresponding area, but the corresponding equipment has not yet fully recovered to normal status. It is necessary to track its status changes, such as by checking the stability of the anomaly propagation probability, to monitor the operation and maintenance process of each device.

[0088] If the current range boundary remains unchanged for at least three consecutive time periods, it indicates that the current risk propagation is stabilizing. For example, the duration of the three consecutive time periods can be 30 minutes per period, and the relative probability error within the three consecutive time periods is ≤5%, indicating that the abnormal propagation process is about to stop. The corresponding area can be output as an isolated segment to clarify the subsequent processing strategy for grid nodes.

[0089] If the number of corresponding grid nodes still changes, it means that the regions corresponding to the three association rules will overlap. It is necessary to conduct comprehensive identification and processing according to the progress of risk in order to achieve fault self-healing migration.

[0090] The target subgraph can represent the isolated area after fault migration, the area of ​​overlapping risks, and the area of ​​risk progression, so as to show the presentation of railway operation and maintenance under different circumstances, and ultimately realize the self-healing of fault migration in railway maintenance.

[0091] In one embodiment of the present invention, step S4 quantifies the response time and status update time of each strategy execution and synchronizes railway operation and maintenance to each grid node to complete the overall strategy deployment.

[0092] The response time mentioned above is the time from when an alarm event is triggered to when maintenance personnel or the system initiates the first handling action, that is, the time from when an alarm event appears on each grid node to when an executable policy is configured on each grid node; the status update time represents the time from the first handling action to when the grid node returns to normal operation. These two times are used to measure the handling status of the corresponding alarm event after the executable policy is used, in order to monitor the handling process of alarm events in railway operation and maintenance scenarios. The alarm handling policy mapped according to these two times can reflect the effect of policy implementation and can be used as reference data for subsequent policy selection.

[0093] Meanwhile, the multidimensional data input into step S4 will include grid node attributes (equipment model, location, rated parameters, etc.), alarm characteristics (alarm level, frequency of occurrence, duration), and resource status (number of maintenance personnel, spare equipment inventory, and maintenance tool configuration). The data at the grid node from the above steps will be summarized to determine the specific situation when the current grid node changes its operating status.

[0094] Therefore, the implementation of step S4 also includes: when the running state of the grid node changes, according to the execution strategy mapped by the grid node, the response time and state update time at the time of execution of the strategy are used as the event correlation between the grid node and the alarm event; and the event correlation corresponding to each grid node is synchronized to the alarm processing strategy.

[0095] After obtaining the output target subgraph, each target subgraph can look up the executable policy corresponding to the current target subgraph in the database, and determine the changes of each grid node after processing according to the response time and status update time of the executable policy during configuration.

[0096] For example, target subgraphs that are completely overlapping can be considered core operations and maintenance subgraphs and need to be processed quickly. For partially overlapping target subgraphs, priority is divided into three levels based on the association rules corresponding to the grid nodes. For example, a subgraph corresponding to three rules is a core operations and maintenance subgraph, a subgraph corresponding to two rules is an associated operations and maintenance subgraph, and a subgraph corresponding to one rule is an external early warning subgraph. This is used to select the corresponding executable strategy. Non-overlapping target subgraphs represent data for which strategies have been configured and are being executed synchronously. For non-overlapping target subgraphs, the strategy used is directly viewed, their stability and isolation are monitored, and the recovery time of the grid nodes is recorded to complete the processing of the overall target subgraph.

[0097] The policies set in the core operations and maintenance sub-graph are used to quickly block faults and prevent them from spreading to critical services. The following actions can be taken: activate backup equipment or links within 3 minutes after an alarm is triggered; suspend non-core services within the sub-graph simultaneously to focus on fault repair; and complete the distribution of fault handling tasks and the response of operations and maintenance personnel within 10 minutes.

[0098] The strategy set by the associated operation and maintenance subgraph is used to link the core operation and maintenance subgraph to prevent the spread of faults to the core operation and maintenance area; its handling actions are to complete remote troubleshooting within 15 minutes; to start monitoring in this area simultaneously when the core operation and maintenance subgraph is handling the fault; to temporarily isolate high-risk nodes when necessary; and to complete the handling in a way that links the core operation and maintenance subgraph.

[0099] The peripheral early warning sub-map focuses on monitoring and early warning without consuming core operation and maintenance resources. It can complete the collaborative processing of multiple grid nodes under railway operation and maintenance by completing alarm information review within 30 minutes, conducting regular inspections to confirm node status, and merging them into the regular maintenance plan when there is no urgent need for handling.

[0100] In one embodiment of the present invention, step S5 will synchronize the status of each grid node in real time based on the progress of the alarm event in the fault migration and self-healing process, so that the development trajectory of the alarm event is described in three dimensions: alarm occurrence time → fault propagation order → strategy execution order. The handling process will sequentially connect the alarm handling strategies of each grid node according to this trajectory. By combining the time combination of alarm occurrence time, response time, and handling completion time, the emergency feasibility of each alarm handling strategy will be analyzed, the optimal path will be selected, and the handling process under the current fault operation and maintenance scheduling mode will be determined.

[0101] The implementation of step S5 also includes: determining the processing duration of the alarm handling strategy at each grid node based on the time combination of alarm occurrence time, response time, and status update time, with the termination condition being that all grid nodes have returned to normal status; the processing duration represents the time from the time recorded when the current alarm event first occurs to the time when the grid node returns to normal status.

[0102] The total processing time of the target subgraph is compared with the processing time, and the alarm handling strategy with the shortest total processing time is used as the output data.

[0103] The output processing flow will record the time consumed by each grid node under the current alarm handling strategy, and output the processing flow for each grid node in the target subgraph according to the recorded time and the total processing time.

[0104] For example, in the grid nodes corresponding to a certain target subgraph, the shortest duration of the core operation and maintenance subgraph node group is 32 minutes; the shortest duration of the associated operation and maintenance subgraph node group is 18 minutes, with a delayed start time of 5 minutes; the shortest duration of the peripheral early warning subgraph node group is 25 minutes, with a delayed start time of 10 minutes. At this time, the total shortest processing time is max(32, 18+5=23, 25+10=35)=35 minutes. The strategies corresponding to these shortest durations are regarded as the current optimal strategy combination, and they are processed in the order of processing as the current output processing flow.

[0105] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention, which are still covered within the protection scope of the present invention.

Claims

1. A method for establishing a digital base station for railway operation and maintenance, characterized in that, include: S1 collects railway operation and maintenance logs and constructs an operation and maintenance topology map composed of multiple grid nodes based on railway lines; S2, guided by the running status of each grid node in the operation and maintenance topology diagram, determines the connection path of each grid node when an alarm event occurs, estimates the association rules for each connection path, and establishes the association rules corresponding to each connection path; S21, For multiple grid nodes on the connection path, select grid nodes with the same business scenario and generate a set of grid nodes; S22, for abnormal nodes in the set of grid nodes that have no direct connection relationship, determine whether the abnormal node satisfies the temporal proximity association. If it does, output the corresponding grid node as a state co-occurrence rule. S23. For the remaining nodes in the set of grid nodes, obtain the upstream and downstream nodes of each grid node in the connection path. Based on the topological relationship of the upstream and downstream nodes, filter the grid nodes that meet the anomaly propagation conditions and output them as state propagation rules. S24. When an upstream node corresponds to multiple abnormal downstream nodes, determine whether there is a temporal proximity association between the downstream nodes. If so, determine that the upstream node is the root cause. Based on the running status of the upstream node, aggregate and combine the corresponding downstream nodes and output them as root cause co-occurrence rules. S25, according to the grid nodes corresponding to the state co-occurrence rule, state propagation rule and root cause co-occurrence rule, store the corresponding data as the association rule corresponding to each connection path; S3, based on the association rules of each connection path, derives the range boundary corresponding to the alarm event, and divides the operation and maintenance topology map into multiple target sub-graphs based on the range boundary; S31, Based on the association rules of each connection path, set the scope boundaries of each association rule; S32, If the boundaries of each association rule completely overlap, output the grid nodes corresponding to each association rule directly as the target subgraph for partitioning; S33, If the boundaries of each association rule partially overlap, the intersection of the boundaries is used to determine the priority. The boundary with the highest priority is used to unify the boundaries of each association rule and output as the target subgraph. S34, if the boundaries of each association rule do not overlap, determine the number of grid nodes in each boundary, perform regional diagnosis on each boundary based on the number of grid nodes, and regard the diagnosed data as the output target subgraph. S4. Based on the multidimensional data of the target subgraph, the execution strategy of each target subgraph is defined. The response time and status update time of each grid node in the target subgraph when executing the strategy are mapped to the alarm handling strategy of each grid node. S5 generates a handling process for railway operation and maintenance scenarios based on the development trajectory of alarm events, according to the alarm handling strategy.

2. The method for establishing a digital base station for railway operation and maintenance according to claim 1, characterized in that, The implementation methods for step S1 include: S11, in response to railway operation and maintenance needs, uses preset query fields to query related fields from railway operation and maintenance logs and determines them as candidate query fields; S12, based on the content corresponding to the candidate query fields, map each candidate query field according to the railway line to form multiple grid nodes; S13: Associate each grid node with its topological connection relationship to construct an operation and maintenance topology diagram for multiple grid nodes.

3. The method for establishing a digital base station for railway operation and maintenance according to claim 2, characterized in that, The implementation of step S13 also includes: S131, upon receiving a railway operation and maintenance request, determine the alarm detection time and alarm confirmation time of the grid node; S132, Based on the alarm detection time and alarm confirmation time of each grid node, perform causal relationship judgment on the grid nodes according to preset logic rules; S133 connects the grid nodes that have completed the judgment and synchronizes them to the operation and maintenance topology diagram to complete the causal relationship of multiple grid nodes under alarm events.

4. The method for establishing a digital base station for railway operation and maintenance according to claim 3, characterized in that, The implementation of step S132 also includes: For any two grid nodes, a preliminary judgment is made based on the timestamps corresponding to the alarm detection time and alarm confirmation time to determine whether the alarm events between the corresponding grid nodes are temporally adjacent. For the grid nodes after the initial judgment, a second judgment is made based on the duration corresponding to the alarm detection time and alarm confirmation time. Based on the judgment results, the causal relationship between the corresponding grid nodes is set.

5. The method for establishing a digital base station for railway operation and maintenance according to claim 4, characterized in that, When selecting mesh nodes that meet the anomaly propagation conditions in step S23, the implementation methods include: Extract the device attributes corresponding to each grid node in the connection path, perform topology relationship query based on the device attributes, and determine the upstream and downstream nodes of the current grid node; Determine the priority of upstream and downstream nodes based on their execution order in the business scenario; The highest priority grid node is selected as the node to be verified. If the node to be verified is abnormal, all reachable grid nodes are marked as candidate propagation nodes starting from the node to be verified. Repeatedly retrieve the highest priority grid node until all grid nodes have been checked; The marked candidate propagation nodes and the abnormal nodes to be checked are considered as outputs of the state propagation rules.

6. The method for establishing a digital base station for railway operation and maintenance according to claim 5, characterized in that, The implementation methods for regional diagnosis in step S34 include: For any grid node within a range boundary, determine the number of grid nodes within the corresponding range boundary. If the number of grid nodes has not changed for at least three consecutive time periods, mark the corresponding location as an isolation segment based on the alarm event corresponding to the current association rule. The probability of abnormal propagation of each grid node within the isolation zone is statistically analyzed. When the probability of abnormal propagation of each grid node is stable, the area where the corresponding grid node is located is regarded as the target sub-graph of the output.

7. The method for establishing a digital base station for railway operation and maintenance according to claim 1, characterized in that, The implementation of step S4 also includes: When the running status of a grid node changes, the response time and status update time at the time of execution of the grid node are used as the event correlation between the grid node and the alarm event, according to the execution strategy mapped to the grid node; the event correlation corresponding to each grid node is synchronized to the alarm handling strategy.

8. A method for establishing a digital base station for railway operation and maintenance according to claim 1, characterized in that, The implementation of step S5 also includes: Based on the time combination of alarm occurrence time, response time, and status update time, and with the termination condition that all grid nodes have returned to normal status, the processing time of the alarm handling strategy in each grid node is determined. The total processing time of the target subgraph is compared with the processing time, and the alarm handling strategy with the shortest total processing time is used as the output data.

Images