Urban rail transit operation and maintenance equipment collaborative diagnosis method, system and platform

Abstract

The application discloses a kind of urban rail transit operation and maintenance equipment collaborative diagnosis method, system and platform, it is related to data management technical field.The method comprises: loading urban rail transit multi-professional operation and maintenance data;To multi-professional operation and maintenance data executes space-time synchronization processing, obtains space-time synchronization operation and maintenance data, constructs cross-professional space-time correlation graph;Load cross-professional knowledge graph;When abnormality appears in multi-professional operation and maintenance data, locate abnormality in the abnormal space-time coordinate in cross-professional space-time correlation graph;With abnormal space-time coordinate as center, according to preset space-time range and space-time coupling coefficient threshold value retrieval graph, obtain associated operation and maintenance data;Associated operation and maintenance data are matched with each historical failure mode, obtain matching result;Output the diagnostic result and collaborative disposal scheme corresponding to matching result.The application realizes the automatic association and collaborative diagnosis of cross-professional data.

Description

Technical Field

[0001] This invention relates to the field of data management technology, specifically to a collaborative diagnostic method, system, and platform for urban rail transit operation and maintenance equipment. Background Technology

[0002] Urban rail transit systems encompass multiple disciplines, including rolling stock, power supply, track maintenance, signaling, and electromechanical systems, each with its own independent monitoring system. However, the inconsistent data formats, asynchronous time bases, and varying spatial positioning methods across these systems make it difficult to automatically correlate and identify cross-disciplinary equipment anomalies occurring within the same time and space, severely hindering collaborative diagnostic capabilities.

[0003] In the prior art, Chinese invention patent CN121388760A discloses a comprehensive monitoring method for urban rail transit based on multi-source heterogeneous data. This method constructs a physical topology map and an operation and maintenance knowledge graph, and introduces a cross-graph attention mechanism for feature fusion to achieve anomaly identification. However, this method focuses on knowledge retrieval triggered by equipment anomalies in the physical topology map and relies on neural networks for feature fusion. It has high requirements for training data, limited interpretability, and fails to address the problems of inconsistent spatiotemporal benchmarks among different professional systems and the difficulty in automatically associating cross-professional data. Furthermore, Chinese invention patent CN120317527B relates to cross-domain collaborative analysis of power equipment faults, but this scheme is applied to the field of power grid dispatching and does not involve the synchronous processing of spatiotemporal labels for multi-professional operation and maintenance data in urban rail transit, nor does it provide a cross-professional data collaborative diagnostic scheme.

[0004] Therefore, there is an urgent need for a method that can unify the spatiotemporal benchmarks of multi-disciplinary data and realize automatic cross-disciplinary data association and collaborative diagnosis. Summary of the Invention

[0005] This invention addresses the technical problems in existing technologies, such as the lack of unified spatiotemporal benchmarks for multi-disciplinary operation and maintenance data, which makes it difficult to automatically associate and identify cross-disciplinary equipment anomalies, the lack of structured expression and quantifiable matching mechanisms for historical fault experience, and the low efficiency of cross-disciplinary collaborative diagnosis. It provides a collaborative diagnosis method, system, and platform for urban rail transit operation and maintenance equipment.

[0006] The technical solution of the present invention to solve the above-mentioned technical problems is as follows:

[0007] In a first aspect, the present invention provides a collaborative diagnostic method for urban rail transit operation and maintenance equipment, comprising: Load multi-disciplinary operation and maintenance data for urban rail transit; Spatiotemporal synchronization processing is performed on the multi-disciplinary operation and maintenance data to obtain spatiotemporal synchronized operation and maintenance data, and a cross-disciplinary spatiotemporal correlation graph is constructed based on the spatiotemporal synchronized operation and maintenance data. The edge weights of the cross-disciplinary spatiotemporal correlation graph are determined by the spatiotemporal coupling coefficient between any two device nodes, and a threshold for the spatiotemporal coupling coefficient is set. Load a cross-disciplinary knowledge graph, which records multiple historical failure modes and cross-disciplinary anomaly spatiotemporal correlation patterns corresponding to each historical failure mode; When an anomaly occurs in the multi-disciplinary operation and maintenance data, locate the anomaly's spatiotemporal coordinates in the cross-disciplinary spatiotemporal correlation map; Centered on the abnormal spatiotemporal coordinates, the cross-professional spatiotemporal correlation map is retrieved according to the preset spatiotemporal range and the spatiotemporal coupling coefficient threshold to obtain the associated operation and maintenance data; The associated operation and maintenance data is matched with each of the historical fault modes. The similarity calculation in the matching process includes the contribution of the spatiotemporal coupling coefficient to obtain the matching result. Output the diagnostic results and collaborative treatment plan corresponding to the matching results.

[0008] Secondly, the present invention provides a collaborative diagnostic system for urban rail transit operation and maintenance equipment, comprising: The data loading module is used to load multi-disciplinary operation and maintenance data for urban rail transit. The graph construction module is used to perform spatiotemporal synchronization processing on the multi-professional operation and maintenance data to obtain spatiotemporal synchronized operation and maintenance data, and to construct a cross-professional spatiotemporal association graph based on the spatiotemporal synchronized operation and maintenance data. The edge weights of the cross-professional spatiotemporal association graph are determined by the spatiotemporal coupling coefficient between any two device nodes, and a threshold for the spatiotemporal coupling coefficient is set. The knowledge loading module is used to load cross-disciplinary knowledge graphs, which record multiple historical fault modes and cross-disciplinary anomaly spatiotemporal correlation patterns corresponding to each historical fault mode. An anomaly localization module is used to locate the anomaly's spatiotemporal coordinates in the cross-professional spatiotemporal correlation map when anomalies occur in the multi-professional operation and maintenance data. The associated retrieval module is used to retrieve the cross-professional spatiotemporal association map based on the abnormal spatiotemporal coordinates, according to a preset spatiotemporal range and the spatiotemporal coupling coefficient threshold, to obtain associated operation and maintenance data. The matching module is used to match the associated operation and maintenance data with each of the historical fault modes. The similarity calculation in the matching process includes the contribution of the spatiotemporal coupling coefficient to obtain the matching result. The result output module is used to output the diagnostic results and collaborative treatment plans corresponding to the matching results.

[0009] Thirdly, the present invention provides a collaborative diagnostic platform for urban rail transit operation and maintenance equipment, wherein the collaborative diagnostic system for urban rail transit operation and maintenance equipment as described in the second aspect is deployed in the urban rail transit operation and maintenance equipment collaborative diagnostic platform.

[0010] The beneficial effects of this invention are: Compared to existing technologies, this invention first performs spatiotemporal synchronization processing on multi-disciplinary operation and maintenance data from urban rail transit vehicles, power supply, track maintenance, signaling, and electromechanical systems. This unifies the independent time bases and spatial positioning methods of each specialty into a standardized spatiotemporal labeling system, solving the inherent defect that prevents automatic association of multi-disciplinary data due to inconsistent formats, asynchronous time, and varying spatial positioning. Secondly, based on the spatiotemporally synchronized operation and maintenance data, a cross-disciplinary spatiotemporal association graph is constructed. Using equipment as nodes and spatiotemporal coupling coefficients as edge weights, the temporal proximity and spatial connectivity between equipment in different specialties are quantified into a computable graph structure, enabling equipment originally scattered across different professional systems to establish quantifiable associations through spatiotemporal dimensions. Thirdly, when an anomaly occurs, the graph is retrieved automatically using the anomaly's spatiotemporal coordinates as the center, within a preset spatiotemporal range and coupling coefficient threshold, to automatically obtain cross-disciplinary operation and maintenance data closely related to the current anomaly in time and space, eliminating the need for manual review of multiple professional systems. Finally, the associated operation and maintenance data is matched with the historical fault modes recorded in the cross-professional knowledge graph. The spatiotemporal coupling coefficient is introduced as a contribution in the similarity calculation, so that the matching results take into account both the similarity of abnormal phenomena and the spatiotemporal correlation strength between devices. The structured diagnostic results and collaborative handling solutions are output, which improves the efficiency and accuracy of cross-professional collaborative diagnosis. Attached Figure Description

[0011] Figure 1 A flowchart illustrating a collaborative diagnostic method for urban rail transit operation and maintenance equipment provided by the present invention; Figure 2 This is a logical schematic diagram of a collaborative diagnostic method for urban rail transit operation and maintenance equipment provided by the present invention; Figure 3 This is a schematic diagram of the structure of a collaborative diagnostic system for urban rail transit operation and maintenance equipment provided by the present invention.

[0012] In the attached diagram, the components represented by each number are as follows: Data loading module 11, graph construction module 12, knowledge loading module 13, anomaly localization module 14, association retrieval module 15, matching module 16, and result output module 17. Detailed Implementation

[0013] Example 1, as Figure 1 , Figure 2 As shown, this embodiment of the invention provides a collaborative diagnostic method for urban rail transit operation and maintenance equipment, including: S10: Load multi-disciplinary operation and maintenance data for urban rail transit; Urban rail transit systems are complex systems encompassing multiple disciplines, each responsible for the operation and maintenance of different equipment and systems. Specifically, the rolling stock discipline deals with core components such as the train's traction, braking, and door systems. Its maintenance data includes train operating status, fault codes, and maintenance records, typically using train numbers and kilometer markers for location. The power supply discipline monitors the operation of power supply equipment such as overhead contact lines, substations, and power cables. Its maintenance data includes voltage and current, switch status, and protection action records, using substation numbers and feeder numbers for location. The track maintenance discipline monitors infrastructure such as rails, ballast, and tunnel structures. Its maintenance data includes track geometry, structural deformation, and crack development, using line sections and mileage for location. The signaling discipline maintains the signaling system, communication system, and automatic train control system. Its maintenance data includes signal status, turnout positions, and communication quality, using station numbers and signal numbers for location. The electromechanical engineering department is responsible for the maintenance of station electromechanical equipment such as ventilation and air conditioning, water supply and drainage, elevators, and platform screen doors. Its operation and maintenance data includes equipment operating parameters, fault alarms, energy consumption data, etc., and the station number and equipment number are used as the location basis.

[0014] Each of the aforementioned specialties has its own independent monitoring system and data acquisition system, with different data formats, sampling frequencies, and encoding methods. The data acquisition frequency for each specialty is set according to the operating characteristics and fault evolution speed of the monitored equipment. For example, the rolling stock specialty deals with high-speed trains, where changes in the traction and braking systems can occur within milliseconds; therefore, the data acquisition frequency is typically set to milliseconds. The electromechanical specialty deals with equipment such as ventilation, air conditioning, and water supply and drainage systems, where changes in status are relatively slow, and the acquisition frequency is typically set to minutes to hours. These frequency differences result in varying data density and granularity across specialties over time, increasing the difficulty of cross-specialty data alignment. The rolling stock specialty uses train numbers as equipment identifiers, while the power supply specialty uses substation numbers as location identifiers. The lack of a unified indexing system between the two makes it difficult to automatically correlate and identify cross-specialty equipment anomalies occurring within the same spatiotemporal range. Therefore, further spatiotemporal synchronization processing is needed to unify the time reference and spatial positioning methods of the data from each specialty.

[0015] This step loads the aforementioned multi-disciplinary operation and maintenance data from various professional systems, including vehicle operation and maintenance data, power supply operation and maintenance data, civil engineering operation and maintenance data, signaling operation and maintenance data, and electromechanical operation and maintenance data. The resulting multi-disciplinary operation and maintenance data will serve as the data source for subsequent spatiotemporal synchronization processing and cross-disciplinary correlation analysis, providing the data input foundation for the entire method.

[0016] S20: Perform spatiotemporal synchronization processing on the multi-professional operation and maintenance data to obtain spatiotemporal synchronized operation and maintenance data, and construct a cross-professional spatiotemporal correlation graph based on the spatiotemporal synchronized operation and maintenance data. The edge weights of the cross-professional spatiotemporal correlation graph are determined by the spatiotemporal coupling coefficient between any two device nodes, and a threshold for the spatiotemporal coupling coefficient is set. Secondly, after loading multi-disciplinary operation and maintenance data, it is necessary to address the core issue of inconsistent spatiotemporal references among the data from different disciplines. Because the vehicle, power supply, track maintenance, signaling, and electromechanical disciplines each use different time references and spatial positioning methods, cross-disciplinary equipment anomalies occurring simultaneously within the same spatial area cannot be automatically correlated and identified. Therefore, this step performs spatiotemporal synchronization processing on the multi-disciplinary operation and maintenance data, unifying the time reference of each discipline's data to the Global Navigation Satellite System (GNSS) time and converting the spatial positioning methods of each discipline into a unified five-level spatial coding system, ensuring that all data are comparable and correlated in both time and space dimensions.

[0017] Specifically, in this embodiment, the multi-professional operation and maintenance data includes vehicle operation and maintenance data, power supply operation and maintenance data, civil engineering operation and maintenance data, signaling operation and maintenance data, and electromechanical operation and maintenance data. The spatiotemporal synchronization processing includes standardization processing and spatiotemporal tag synchronization processing. The standardization process includes: converting the original units of each physical quantity in each professional operation and maintenance data into preset standard international units; mapping the original coding system of each professional operation and maintenance data into a unified coding dictionary according to preset standardization mapping rules; unifying the sampling frequency of each professional operation and maintenance data to a preset standard time granularity; and converting the different data formats of each professional operation and maintenance data into a preset unified data structure model. The spatiotemporal tag synchronization processing includes: using the Global Navigation Satellite System as a unified time source to perform synchronous calibration on the data acquisition time of each professional system to obtain a unified time tag; constructing a five-level spatial coding system and converting the original spatial positioning methods of each professional system into the five-level spatial coding system through a preset spatial mapping relationship table to obtain a unified spatial tag; and adding the unified time tag and the unified spatial tag to the standardized data to obtain the spatiotemporal synchronized operation and maintenance data.

[0018] The five-level spatial coding system includes line number, station number, up and down direction, relative mileage, and equipment level. For the vehicle professional operation and maintenance data, the train number and kilometer marker are mapped to the five-level spatial coding system through the preset spatial mapping relationship table; For the power supply operation and maintenance data, the substation number and feeder number are mapped to the five-level spatial coding system through the preset spatial mapping relationship table; For the aforementioned engineering maintenance data, the line sections and mileages are mapped to the five-level spatial coding system through the preset spatial mapping relationship table; For the signaling professional operation and maintenance data, the station and signal number are mapped to the five-level spatial coding system through the preset spatial mapping relationship table.

[0019] In this embodiment, the multi-disciplinary operation and maintenance data includes vehicle operation and maintenance data, power supply operation and maintenance data, civil engineering operation and maintenance data, signaling operation and maintenance data, and electromechanical operation and maintenance data.

[0020] Spatiotemporal synchronization processing includes two parts: normalization processing and spatiotemporal tag synchronization processing. Standardization processing includes the following: First, converting the original units of physical quantities in the maintenance data from various specialties to preset standard international units. Different specialties use different units of physical quantities; for example, vehicle maintenance uses megapascals, power supply uses kilopascals, and electromechanical uses bar. Converting to a unified standard international unit eliminates the impact of unit differences on data fusion. Second, mapping the original coding systems of each specialty's maintenance data to a unified coding dictionary according to preset standardized mapping rules. Different specialties use different coding methods for information such as equipment status and fault types; for example, fault codes are represented by letters and numbers in vehicle maintenance and by pure numbers in power supply maintenance. Mapping to a unified coding dictionary ensures consistency in the way different specialties express the same type of information. Third, unifying the sampling frequency of maintenance data from various specialties to a preset standard time granularity. Different specialties have different data sampling frequencies; for example, vehicle maintenance data is collected at the millisecond level, while civil engineering data is collected at the hour level. Unifying to the same time granularity through resampling or interpolation methods makes the data comparable across time dimensions. The preset standard time granularity is set comprehensively based on the distribution range of data sampling frequency in various professional systems of urban rail transit and the time alignment accuracy requirements of cross-professional correlation analysis. For example, it can be set to 1 second. Fourth, the different data formats of operation and maintenance data of each professional system are converted into a preset unified data structure model. The data storage formats of each professional system are different. For example, the vehicle system uses a time-series database format, while the power supply system uses a relational database format. By converting it into a unified data structure, it is easier to perform subsequent correlation storage and retrieval.

[0021] The spatiotemporal tag synchronization process includes the following: First, using the Global Navigation Satellite System (GNSS) as a unified time source, the data acquisition time of each specialized system is synchronized and calibrated to obtain a unified time tag. Since the internal clocks of each specialized system may have discrepancies, calibration is performed using GNSS time synchronization to ensure that all data is aligned with the same benchmark in the time dimension.

[0022] Second, a five-level spatial coding system is constructed, and the original spatial positioning methods of each professional system are converted into the five-level spatial coding system through a preset spatial mapping table to obtain a unified spatial label. The five-level spatial coding system includes five levels: line number, station number, up / down direction, relative mileage, and equipment level. It locates the precise position of equipment in the three-dimensional space of urban rail transit from coarse to fine, and can uniquely determine the precise spatial position of any piece of equipment in the urban rail transit system. Among them, the line number identifies the urban rail transit line to which the equipment belongs; the station number identifies the station where the equipment is located or nearby; the up / down direction indicates whether the equipment is located on the up or down line; the relative mileage indicates the specific kilometer marker position of the equipment on the line, with an accuracy of meters; and the equipment level indicates the category of the equipment in the system hierarchy, which may include four levels: line level, station level, section level, and equipment level.

[0023] The original spatial positioning methods of each professional system are uniformly converted into this five-level spatial coding system through a pre-set spatial mapping table, obtaining standardized spatial labels and realizing unified spatial association and retrieval of cross-professional data. Specifically, the pre-set spatial mapping table is used to uniformly convert the original spatial positioning methods of each professional system into the five-level spatial coding system. The mapping table includes the mapping rules between train numbers and kilometer markers for the rolling stock system, substation numbers and feeder numbers for the power supply system, line sections and mileage for the track maintenance system, and station and signal numbers for the signaling system, and the corresponding five-level spatial coding system.

[0024] Preferably, for vehicle maintenance data, a preset spatial mapping table maps train numbers and kilometer markers to a five-level spatial coding system. Raw vehicle data is located using train numbers and kilometer markers, such as "Train T1234, K15+200". The preset spatial mapping table converts the train's current location into fields such as line number, station number, and relative mileage in the five-level spatial coding system. For power supply maintenance data, a preset spatial mapping table maps substation numbers and feeder numbers to a five-level spatial coding system. Raw power supply data is located using substation numbers and feeder numbers, such as "Substation S05, Feeder F12". The preset spatial mapping table converts the equipment location into a five-level spatial coding system. For track maintenance data, a preset spatial mapping table maps track sections and mileage to a five-level spatial coding system. Raw track maintenance data is located using track sections and mileage, such as "Upstream section K10+000 to K10+500". The preset spatial mapping table converts this into a five-level spatial coding system. For signaling maintenance data, a pre-defined spatial mapping table maps station and signal controller numbers to a five-level spatial coding system. Original signaling data is located by station and signal controller numbers, such as "Station Z03, Signal Controller X05," and is converted to a five-level spatial coding system using the pre-defined spatial mapping table. Electromechanical maintenance data is located by station and equipment numbers, and is also converted to a five-level spatial coding system using the pre-defined spatial mapping table.

[0025] Through the above spatiotemporal synchronization processing, data from various disciplines can obtain unified, alignable identifiers in both time and space dimensions, providing a foundation for the subsequent construction of cross-disciplinary spatiotemporal correlation maps and cross-disciplinary anomaly retrieval.

[0026] Third, add unified time tags and unified spatial tags to the standardized data to obtain spatiotemporally synchronized operation and maintenance data.

[0027] Furthermore, based on this, a cross-disciplinary spatiotemporal correlation graph is constructed, using devices as nodes and the spatiotemporal coupling coefficient between any two device nodes as edge weights. The spatiotemporal coupling coefficient is a comprehensive indicator that quantifies the strength of the correlation between two device nodes in the temporal and spatial dimensions. It is obtained by weighted fusion of the temporal dimension correlation weight determined by the time deviation tolerance and the spatial dimension correlation weight determined by the spatial topological connection distance. A higher spatiotemporal coupling coefficient indicates a tighter correlation between the two devices in the spatiotemporal dimension, and a greater likelihood that the other device will also be in an abnormal state when one device malfunctions. A threshold for the spatiotemporal coupling coefficient is also set to filter device node pairs with significant spatiotemporal correlations during subsequent retrieval. Through the construction of this graph, devices originally scattered across different professional systems establish quantifiable correlations in the spatiotemporal dimension, providing a structured data foundation for the automatic correlation identification of cross-disciplinary anomalies.

[0028] Specifically, in this embodiment, the spatiotemporal coupling coefficient is determined in the following way: Obtain the time skew tolerance and spatial topology connection distance between any two device nodes; The time dimension association weight is determined based on the time deviation tolerance, and the spatial dimension association weight is determined based on the spatial topology connection distance and the type of equipment along the urban rail transit line. The time-dimensional association weights and the spatial-dimensional association weights are weighted and fused to obtain the spatiotemporal coupling coefficient.

[0029] First, obtain the time deviation tolerance and spatial topology connection distance between any two device nodes. The time deviation tolerance refers to the maximum permissible time deviation between the data acquisition times of two device nodes, determined based on the time synchronization accuracy and data type of each professional system. For example, for the vehicle and power supply specialties involving high-speed train operation, with data acquisition frequencies in the millisecond range, the time deviation tolerance can be set to 100 milliseconds; for the civil engineering and electromechanical specialties, with data acquisition frequencies in the minute or hour range, the time deviation tolerance can be set to 1 minute or 10 minutes. The time deviation tolerance reflects the maximum permissible deviation in the time dimension that can be considered to occur simultaneously in the data of two device nodes when performing cross-professional data association.

[0030] Spatial topology connection distance refers to the length of the connection path or spatial proximity of two equipment nodes in the physical topology along an urban rail transit line, calculated based on the physical connection relationship or spatial location between the equipment. For example, the spatial topology connection distance between two devices within the same substation is shorter, while the spatial topology connection distance between two devices located at adjacent stations is longer; the topology distance between two devices directly connected by cables is shorter than the distance between two devices indirectly connected by multiple intermediate nodes.

[0031] Secondly, the time dimension correlation weight is determined based on the time deviation tolerance. The time dimension correlation weight is determined by the ratio of the time deviation between the data acquisition times of the two device nodes to the time deviation tolerance; the smaller the time deviation, the higher the weight. Specifically, the calculation method is: Time dimension correlation weight = 1 - Time deviation / Time deviation tolerance. When the time deviation exceeds the time deviation tolerance, it indicates that the difference in the time dimension between the data acquisition times of the two device nodes has exceeded the allowable range, and there is no effective time synchronization between them; the weight is 0. For example, if the time deviation tolerance of the two device nodes is 100 milliseconds and the actual time deviation is 20 milliseconds, then the time dimension correlation weight is 1 - 20 / 100 = 0.8; if the actual time deviation is 120 milliseconds, exceeding the tolerance, then the weight is 0. This time dimension correlation weight reflects the degree of synchronization between the two device nodes in the time dimension; the smaller the time deviation, the closer they are in time, and the higher the reliability of the cross-disciplinary correlation.

[0032] Furthermore, spatial dimension association weights are determined based on the spatial topological connection distance and the equipment types along the urban rail transit line. The spatial dimension association weights are determined by combining the spatial topological connection distance between two equipment nodes with the fault propagation probability between equipment types; the closer the distance and the higher the fault propagation probability between equipment types, the higher the weight. Specifically, the calculation method is as follows: First, spatial proximity is calculated based on the spatial topological connection distance between two equipment nodes. Spatial proximity is negatively correlated with spatial topological connection distance; for example, spatial proximity = 1 / 1 + the normalized value of spatial topological connection distance. Then, a fault propagation probability coefficient is determined based on the equipment types along the urban rail transit line. The fault propagation probability coefficient represents the probability that when one equipment type fails, another equipment type will be affected and fail as a result. This fault propagation probability coefficient can be obtained through historical fault statistics. Spatial dimension association weight = spatial proximity × fault propagation probability coefficient.

[0033] For example, the overhead contact line and the pantograph are in direct physical contact, resulting in a high probability of fault propagation. If the spatial topological connection distance between the two is short, the spatial dimension correlation weight is high. The air conditioning system and the traction system are not in direct physical contact, resulting in a low probability of fault propagation. Even if the spatial distance is short, the spatial dimension correlation weight is low.

[0034] Finally, a weighted fusion is performed on the temporal and spatial correlation weights to obtain the spatiotemporal coupling coefficient. The formula for calculating the spatiotemporal coupling coefficient is: Spatiotemporal Coupling ...

[0035] Through the above calculations, each edge in the cross-disciplinary spatiotemporal correlation graph carries a quantized spatiotemporal coupling coefficient, which can provide a basis for subsequent retrieval of associated operation and maintenance data based on the threshold of the spatiotemporal coupling coefficient.

[0036] Specifically, in this embodiment, constructing a cross-professional spatiotemporal correlation map based on the spatiotemporal synchronized operation and maintenance data includes: The spatiotemporal synchronization operation and maintenance data is stored in association according to the time and space dimensions. The cross-professional spatiotemporal association graph is constructed with devices as nodes and the spatiotemporal coupling coefficient between any two device nodes as edge weights.

[0037] First, the spatiotemporally synchronized operation and maintenance data are stored together using unified time tags and unified spatial tags. After spatiotemporal synchronization, each piece of operation and maintenance data carries a unified time tag and a unified spatial tag. The unified time tag is a timestamp calibrated by the Global Navigation Satellite System, ensuring a unified alignment benchmark for data from different disciplines in the time dimension; the unified spatial tag is a spatial location code encoded using a five-level spatial coding system, ensuring a unified positioning benchmark for data from different disciplines in the spatial dimension.

[0038] The above data is stored in a structured manner in the database according to time and spatial tags, and a data organization method with devices as the basic unit is established. This ensures that all historical operation and maintenance data of each device are arranged in order of time and spatial location, which facilitates subsequent map construction and retrieval operations.

[0039] Furthermore, using equipment as nodes and the spatiotemporal coupling coefficient between any two equipment nodes as edge weights, a cross-disciplinary spatiotemporal association graph is constructed. Each node in the graph represents a piece of equipment in the urban rail transit system, and the node attributes include the equipment's unified spatial label, its affiliated discipline, equipment type, and other information. Each edge in the graph connects two equipment nodes, and the weight of the edge is the spatiotemporal coupling coefficient between these two equipment nodes. As mentioned earlier, the spatiotemporal coupling coefficient comprehensively reflects the degree of synchronization between two devices in the time dimension and their proximity in the spatial dimension, as well as their failure propagation probability. When the spatiotemporal coupling coefficient between two equipment nodes is greater than zero, it indicates that there is a quantifiable spatiotemporal association between them; when the spatiotemporal coupling coefficient is zero, it indicates that there is no effective association between them, and no edge connection is established in the graph.

[0040] Through the above construction method, the cross-disciplinary spatiotemporal correlation graph places all professional equipment in a unified graph structure. Equipment originally scattered across different professional systems such as vehicle, power supply, track maintenance, signaling, and electromechanical systems establishes cross-disciplinary correlation edges through spatiotemporal coupling coefficients. This cross-disciplinary spatiotemporal correlation graph provides a directly usable graph data structure for subsequent retrieval of spatiotemporally adjacent related equipment starting from abnormal equipment nodes, transforming the cross-disciplinary data retrieval problem into a neighborhood search problem on the graph.

[0041] S30: Load the cross-professional knowledge graph, which records multiple historical fault modes and cross-professional anomaly spatiotemporal correlation patterns corresponding to each historical fault mode; Furthermore, after constructing the cross-disciplinary spatiotemporal correlation graph, it is also necessary to load the cross-disciplinary knowledge graph. The cross-disciplinary spatiotemporal correlation graph reflects the spatiotemporal relationships between various devices, but it does not contain historical fault experience knowledge. When an anomaly occurs, it is not only necessary to retrieve the spatiotemporally correlated cross-disciplinary device data, but also to determine whether the current anomaly spatiotemporal correlation pattern matches historically occurring fault patterns, so as to quickly locate the root cause and formulate a response plan by drawing on historical experience.

[0042] Cross-disciplinary knowledge graphs are structured fault knowledge bases accumulated during long-term operation and maintenance. They record multiple historical fault modes and corresponding cross-disciplinary anomaly spatiotemporal correlation patterns for each historical fault mode. Each historical fault mode includes the following information: fault type identifier, list of involved disciplines, list of involved equipment nodes, chronological order of occurrence of anomalies in each equipment, and spatial propagation path between anomalies in each equipment. Cross-disciplinary anomaly spatiotemporal correlation patterns are a formalized expression of the temporal and spatial relationships between cross-disciplinary equipment anomalies in historical fault events. They record the temporal order, spatial distribution, and propagation direction of anomalies in each professional equipment under this fault mode.

[0043] For example, in a historical fault event, a voltage dip occurred in the power supply overhead contact line, followed by an anomaly in the signaling system, and then an alarm was triggered by the rolling stock traction system. The corresponding cross-disciplinary anomaly spatiotemporal correlation pattern records the temporal sequence of these three equipment anomalies and their spatial propagation direction—from the power supply equipment to the signaling equipment and then to the rolling stock equipment. When the spatiotemporal correlation pattern of the current anomaly matches that of the historical fault pattern, it can be inferred that the current anomaly may share the same root cause as the historical fault, thus allowing direct access to the diagnostic results and handling solutions corresponding to that historical fault pattern.

[0044] It should be noted that the historical fault mode data in the cross-knowledge knowledge graph originates from historical fault reports and collaborative handling records accumulated by urban rail transit operation and maintenance departments over long-term operations. This data has been structured and stored in the knowledge graph according to a unified spatiotemporal association pattern format. The cross-knowledge knowledge graph and the cross-knowledge spatiotemporal association graph together constitute the two knowledge foundations of this method: the former provides historical experience knowledge, and the latter provides the spatiotemporal relationships between current equipment. After loading the cross-knowledge knowledge graph, the subsequent anomaly localization, association retrieval, and fault mode matching stages can begin.

[0045] S40: When an anomaly occurs in the multi-professional operation and maintenance data, locate the anomaly's spatiotemporal coordinates in the cross-professional spatiotemporal correlation map; Specifically, after loading multi-disciplinary operation and maintenance data and completing spatiotemporal synchronization processing, real-time monitoring data from each specialty continuously flows into the system. Each specialty system analyzes the data stream in real time based on its own monitoring thresholds and anomaly judgment rules. When a specialty system detects that the operating parameters of a device exceed the normal range or trigger a preset alarm condition, it determines that the device has malfunctioned. For example, if the power supply specialty system detects a sudden increase in feeder current in a substation exceeding the protection setting, or if the vehicle specialty system detects that the brake cylinder pressure of a train is below the lower limit threshold, both will trigger an abnormal event. When an abnormal event is triggered, the abnormal device identifier, the time of the abnormality, and the spatial location of the abnormal device are recorded.

[0046] Next, the spatiotemporal coordinates of the anomaly are located within the cross-disciplinary spatiotemporal correlation map. Anomaly spatiotemporal coordinates are precise location identifiers of anomaly events in both time and space dimensions. In the time dimension, the anomaly spatiotemporal coordinates include the time of anomaly occurrence and the duration window of the anomaly. The time of anomaly occurrence is the starting time of anomaly determination, and the duration window is the time length from the start of the anomaly to its recovery or the current time. In the spatial dimension, the anomaly spatiotemporal coordinates are determined by the unified spatial label of the anomalous equipment in the five-level spatial coding system, including information such as line number, station number, up / down direction, relative mileage, and equipment level. Combining the above information from the time and spatial dimensions yields the anomaly spatiotemporal coordinates within the cross-disciplinary spatiotemporal correlation map.

[0047] The abnormal spatiotemporal coordinates will serve as the spatiotemporal anchor point for subsequent retrieval of related operation and maintenance data. That is, using these abnormal spatiotemporal coordinates as the center, the search will be conducted within the cross-disciplinary spatiotemporal correlation map to find spatiotemporally adjacent cross-disciplinary equipment and their corresponding operation and maintenance data. By accurately locating the abnormal spatiotemporal coordinates, it can be ensured that the scope of subsequent correlation retrieval is expanded based on the actual spatiotemporal location of the current abnormal event.

[0048] S50: Using the abnormal spatiotemporal coordinates as the center, retrieve the cross-professional spatiotemporal correlation map according to the preset spatiotemporal range and the spatiotemporal coupling coefficient threshold to obtain the associated operation and maintenance data; Specifically, in this embodiment, the cross-professional spatiotemporal correlation map is retrieved based on the abnormal spatiotemporal coordinates as the center, according to a preset spatiotemporal range and the spatiotemporal coupling coefficient threshold, to obtain the associated operation and maintenance data, including: A temporal neighborhood is obtained by expanding the abnormal time interval in the abnormal spatiotemporal coordinates according to a preset time radius threshold, and a spatial neighborhood is obtained by expanding the abnormal spatial range in the abnormal spatiotemporal coordinates according to a preset spatial radius threshold. Merge the temporal neighborhood and the spatial neighborhood, and extract the device node pairs in which the spatiotemporal coupling coefficient is greater than or equal to the spatiotemporal coupling coefficient threshold, and determine them as the same spatiotemporal neighborhood; The associated operation and maintenance data is obtained by retrieving all device nodes falling within the same spatiotemporal neighborhood and the corresponding operation and maintenance data of each device node from the cross-professional spatiotemporal correlation map.

[0049] First, a time neighborhood is obtained by expanding the abnormal time interval in the abnormal spatiotemporal coordinates according to a preset time radius threshold. The abnormal time interval in the abnormal spatiotemporal coordinates records the duration of the abnormality from its start to the current moment. The preset time radius threshold is the time range expanded to both sides of the start and end points of the abnormal time interval, for example, it can be set to 5 minutes before and after the time of the abnormality. The abnormal time interval plus the expanded time range before and after it together constitute the time neighborhood. This time neighborhood is used to limit the scope of correlation retrieval in the time dimension, only retrieving data records that occurred within this time neighborhood, avoiding the introduction of irrelevant data due to an excessively large retrieval scope. Specifically, the preset time radius threshold can be set according to the type of abnormality and the characteristics of the equipment. For example, a shorter time radius can be set for transient electrical faults, and a longer time radius can be set for progressive mechanical faults.

[0050] Simultaneously, a spatial neighborhood is obtained by expanding the abnormal spatial range in the abnormal spatiotemporal coordinates according to a preset spatial radius threshold. The abnormal spatial range is the spatial location of the abnormal equipment, determined by the line number, station number, and relative mileage in the five-level spatial coding system. The preset spatial radius threshold is the spatial range expanding to both sides of the abnormal equipment, determined based on the spatiotemporal propagation speed of urban rail transit equipment faults and statistical analysis of historical fault data. For example, it can be set to a range of 500 meters before and after the abnormal equipment or between two adjacent stations. The abnormal spatial location plus the expanded spatial range before and after it together constitute the spatial neighborhood. This spatial neighborhood is used to limit the scope of the correlation search in the spatial dimension, only searching for equipment installed or operating within this spatial neighborhood, avoiding the introduction of unrelated equipment that is spatially far away from the abnormal equipment due to an excessively large search scope.

[0051] Furthermore, the temporal and spatial neighborhoods are merged, and device node pairs with a spatiotemporal coupling coefficient greater than or equal to a threshold are extracted and identified as belonging to the same spatiotemporal neighborhood. Merging refers to taking the intersection of the temporal and spatial neighborhoods, i.e., devices and data records that simultaneously satisfy both temporal and spatial conditions. Within this intersection, the spatiotemporal coupling coefficient between any two device nodes is calculated, and device node pairs with a spatiotemporal coupling coefficient greater than or equal to a preset threshold are retained, thus identifying these pairs as nodes within the same spatiotemporal neighborhood. The spatiotemporal coupling coefficient threshold controls the strictness of the association retrieval. It is determined based on the actual spatiotemporal coupling coefficient distribution of cross-disciplinary abnormal device pairs in historical fault events. A higher threshold means the retrieval results only include device pairs with close spatiotemporal associations; a lower threshold means the retrieval results include more device pairs with weaker associations. For example, a spatiotemporal coupling coefficient threshold of 0.6 can be set to retain only device node pairs with a spatiotemporal coupling coefficient greater than or equal to 0.6.

[0052] Finally, the cross-disciplinary spatiotemporal correlation graph retrieves all device nodes falling within the same spatiotemporal neighborhood and their corresponding professional operation and maintenance data to obtain the associated operation and maintenance data. The cross-disciplinary spatiotemporal correlation graph stores all devices and their relationships, with devices as nodes and spatiotemporal coupling coefficients as edge weights. After identifying device node pairs within the same spatiotemporal neighborhood, the graph extracts the devices represented by the nodes and all corresponding professional operation and maintenance data, including data from vehicle engineering, power supply, civil engineering, signaling, and electromechanical engineering. This data constitutes the associated operation and maintenance data, covering cross-disciplinary device status information closely related to the current anomaly in both time and space.

[0053] Through the above retrieval process, without the need for manual consultation of multiple professional systems, cross-professional operation and maintenance data with significant spatiotemporal correlation with the current anomaly can be automatically retrieved from the cross-professional spatiotemporal correlation map, providing structured input for subsequent fault mode matching.

[0054] S60: Match the associated operation and maintenance data with each of the historical fault modes. The similarity calculation in the matching process includes the contribution of the spatiotemporal coupling coefficient to obtain the matching result. Furthermore, after obtaining the associated maintenance data through cross-disciplinary spatiotemporal correlation graph retrieval, it is necessary to determine whether the spatiotemporal pattern of the current anomaly and its associated data matches the historical fault patterns recorded in the cross-disciplinary knowledge graph. The associated maintenance data contains cross-disciplinary equipment status information closely related to the current anomaly in time and space, reflecting the cross-disciplinary anomaly spatiotemporal correlation pattern at the time the current anomaly occurred. Each historical fault pattern stored in the cross-disciplinary knowledge graph contains a corresponding cross-disciplinary anomaly spatiotemporal correlation pattern, recording the temporal sequence, spatial distribution, and propagation path of anomalies in various professional equipment during that historical fault event. Matching the associated maintenance data corresponding to the current anomaly with each historical fault pattern allows determination of whether the current anomaly shares a similar cross-disciplinary anomaly spatiotemporal correlation pattern with a historical fault, thus enabling rapid diagnosis based on historical experience.

[0055] Specifically, in this embodiment, the associated operation and maintenance data is matched with each of the historical fault modes. The similarity calculation in the matching process includes a contribution term of the spatiotemporal coupling coefficient, including: Extract the temporal sequence and spatial relative position of anomalies in various professional data from the associated operation and maintenance data to form a spatiotemporal association pattern of anomalies to be matched. Obtain the spatiotemporal coupling coefficient between any two device nodes in the associated operation and maintenance data, and calculate the benchmark similarity between the cross-professional spatiotemporal correlation pattern of the abnormal spatiotemporal correlation pattern to be matched and the cross-professional abnormal spatiotemporal correlation pattern of each historical fault pattern. The weighted values ​​of the baseline similarity and the spatiotemporal coupling coefficient are superimposed to obtain the final similarity corresponding to each of the historical fault modes; Extract historical fault patterns whose final similarity is greater than a preset similarity threshold to obtain the matching result.

[0056] First, the temporal sequence and spatial relative position of anomalies across various disciplines are extracted from the associated maintenance data to form a spatiotemporal correlation pattern for the anomaly to be matched. The associated maintenance data contains cross-discipline equipment status information closely related to the current anomaly in time and space. From this data, the equipment nodes that experienced anomalies are extracted, and the time of each anomaly is recorded and arranged chronologically to form an anomaly time series. Simultaneously, the spatial location of each anomalous equipment node in the five-level spatial coding system is recorded, and the spatial relative positional relationship between the anomalous equipment nodes is calculated, such as whether they are located at the same station, the same line section, or in corresponding up / down directions. The temporal sequence and spatial relative positional relationship are combined to form the spatiotemporal correlation pattern for the anomaly to be matched. This spatiotemporal correlation pattern is a quantitative expression of the spatiotemporal structure of the current anomaly event across disciplines.

[0057] Secondly, the spatiotemporal coupling coefficient between any two device nodes in the associated operation and maintenance data is obtained, and the baseline similarity between the spatiotemporal correlation pattern of the anomaly to be matched and the cross-disciplinary spatiotemporal correlation patterns of each historical fault pattern is calculated. For each historical fault pattern in the cross-disciplinary knowledge graph, its corresponding cross-disciplinary spatiotemporal correlation pattern is extracted, including the temporal order and spatial relative position of each abnormal device node in the historical fault. The spatiotemporal correlation pattern of the anomaly to be matched is compared with the cross-disciplinary spatiotemporal correlation pattern of the historical fault pattern, and the baseline similarity between the two is calculated. The baseline similarity can be calculated using a weighted combination of time series matching algorithm and spatial topology matching algorithm. In the time dimension, the occurrence order of each abnormal event in the two patterns is compared to see if they are consistent, and the time interval between each adjacent event is similar. In the spatial dimension, the spatial distribution of each abnormal device node in the two patterns is compared to see if they are similar, and the fault propagation direction is consistent. The baseline similarity is obtained by weighted summation of the time dimension similarity and the spatial dimension similarity.

[0058] Furthermore, the weighted values ​​of the baseline similarity and the spatiotemporal coupling coefficient are superimposed to obtain the final similarity corresponding to each historical fault mode. The spatiotemporal coupling coefficient reflects the actual temporal and spatial correlation strength between abnormal device nodes in the current associated maintenance data. In the current abnormal mode, the higher the spatiotemporal coupling coefficient between two device nodes, the closer the spatiotemporal correlation between them, and the higher the confidence level of the matching result of these two nodes should be obtained during the matching process. Specifically, for each pair of device nodes in the spatiotemporal correlation mode to be matched, its spatiotemporal coupling coefficient is obtained, the average value of the spatiotemporal coupling coefficients of all device node pairs in the current mode is calculated, and this average value is multiplied by a preset weight coefficient and superimposed on the baseline similarity. The formula for calculating the final similarity is: Final similarity = Baseline similarity + Average value of spatiotemporal coupling coefficient × Weight coefficient. By introducing the contribution of the spatiotemporal coupling coefficient, the matching result not only considers the temporal order and spatial layout similarity of the abnormal phenomena, but also the strength of the actual spatiotemporal correlation between each device node in the current abnormality. The higher the spatiotemporal coupling coefficient, the more reliable the cross-professional correlation of the current abnormality, and the higher the confidence level should be obtained in the matching.

[0059] Finally, historical fault patterns with a final similarity greater than a preset similarity threshold are extracted to obtain matching results. The preset similarity threshold can be set based on the accuracy requirements and mismatch tolerance for historical fault pattern matching; for example, it can be set to 0.75. Historical fault patterns with a final similarity greater than this preset similarity threshold are considered to match the current anomaly, and this pattern is output as the matching result. If multiple matching results exist, they can be sorted from highest to lowest final similarity, with the historical fault pattern with the highest similarity being output first. If the final similarity of all historical fault patterns does not reach the threshold, the matching result is empty, indicating that the current anomaly belongs to a new, unrecorded pattern, and the process can proceed to the storage of new anomaly patterns.

[0060] Through the above matching process, historical fault experience can be used to quickly identify and diagnose current cross-disciplinary anomalies. At the same time, by introducing the spatiotemporal coupling coefficient, the sensitivity of the matching results to the cross-disciplinary correlation strength is improved.

[0061] S70: Output the diagnostic results and collaborative treatment plan corresponding to the matching results.

[0062] Finally, after matching the associated operation and maintenance data with historical failure modes, the matching results need to be transformed into diagnostic conclusions and handling instructions that can be directly used by operation and maintenance personnel.

[0063] Specifically, the execution steps of this embodiment further include: When a matching target historical fault pattern exists in the matching results, a fault propagation link diagram is constructed based on the cross-professional anomaly spatiotemporal correlation pattern of the target historical fault pattern. The fault propagation link diagram includes the fault initiation professional identifier, the fault propagation path, the list of affected professional disciplines, and the temporal relationship between each node. When no matching target historical fault mode is found in the matching results, the abnormal spatiotemporal coordinates, the associated operation and maintenance data, and the spatiotemporal sequence of each professional anomaly in the associated operation and maintenance data are combined into a new anomaly mode and added to the cross-knowledge knowledge graph.

[0064] When a matching historical failure mode is found in the matching results, the diagnostic results and collaborative handling plan associated with that historical failure mode are retrieved from the cross-knowledge knowledge graph. The diagnostic results include the root cause of the failure, the failure type, the scope of involved specialties and equipment, and the reliability rating of that failure mode in historical cases. The collaborative handling plan is a pre-defined handling process for that failure mode, including the operational steps to be performed by each specialty, the order of handling, the collaboration methods between specialties, the list of required tools and spare parts, and safety precautions during the handling process. This information is then structured and output to the operations and maintenance management interface to guide the various specialties in collaboratively carrying out failure handling work.

[0065] Simultaneously, a fault propagation link graph is constructed based on the cross-disciplinary anomaly spatiotemporal correlation pattern of the target historical fault mode, presenting the fault propagation path in a visual form. The fault propagation link graph includes the fault initiation discipline identifier, the fault propagation path, the list of affected disciplines, and the temporal relationship between each node. If no matching target historical fault mode is found in the matching results, it indicates that the current anomaly belongs to a new, unrecorded mode. The system combines the anomaly spatiotemporal coordinates corresponding to the current anomaly, the associated operation and maintenance data, and the spatiotemporal sequence of anomalies in each discipline within the associated operation and maintenance data into a new anomaly mode, adding it to the cross-disciplinary knowledge graph for reference when similar anomalies occur in the future.

[0066] Through the aforementioned output mechanism, the results of cross-disciplinary collaborative diagnosis can be presented to operations and maintenance personnel in the form of structured diagnostic reports, handling plans, and visualized process diagrams. This shortens the time from anomaly discovery to handling decisions and improves the efficiency of cross-disciplinary collaborative handling. Simultaneously, the new knowledge storage mechanism allows the cross-disciplinary knowledge graph to be continuously updated and expanded as operational and maintenance experience accumulates.

[0067] In summary, the embodiments of this application have at least the following technical effects: This invention first establishes a unified spatiotemporal benchmark for multi-disciplinary operation and maintenance data. Through standardization processing and spatiotemporal tag synchronization, heterogeneous data from various disciplines such as vehicle, power supply, civil engineering, signaling, and electromechanical engineering are converted into synchronized data with unified time tags and five-level spatial coding, providing a data foundation for cross-disciplinary correlation analysis. Secondly, it constructs a cross-disciplinary spatiotemporal correlation graph with equipment as nodes and spatiotemporal coupling coefficients as edge weights. Time deviation tolerance and spatial topological connection distance are weighted and fused into edge weights to quantitatively characterize the correlation strength between different professional equipment in the spatiotemporal dimension, providing a quantitative basis for cross-disciplinary equipment anomaly retrieval.

[0068] Secondly, when an anomaly occurs, the system automatically retrieves related operational data from the knowledge graph based on the anomaly's spatiotemporal coordinates, within a preset spatiotemporal range and a spatiotemporal coupling coefficient threshold. This enables automatic cross-disciplinary data retrieval, replacing manual cross-system searches. Finally, the related data is matched with historical fault patterns. The spatiotemporal coupling coefficient is incorporated as a contribution factor in the similarity calculation. Upon successful matching, a diagnostic result containing a fault propagation chain diagram and a collaborative handling plan are output. If a match fails, the new pattern is added to the knowledge graph to achieve continuous knowledge accumulation and form a closed-loop optimization mechanism.

[0069] Example 2, as Figure 3 As shown, based on the same inventive concept as the collaborative diagnostic method for urban rail transit operation and maintenance equipment provided in Embodiment 1, this embodiment of the invention also provides a collaborative diagnostic system for urban rail transit operation and maintenance equipment, including: Data loading module 11 is used to load multi-disciplinary operation and maintenance data of urban rail transit; The graph construction module 12 is used to perform spatiotemporal synchronization processing on the multi-professional operation and maintenance data to obtain spatiotemporal synchronized operation and maintenance data, and to construct a cross-professional spatiotemporal association graph based on the spatiotemporal synchronized operation and maintenance data. The edge weights of the cross-professional spatiotemporal association graph are determined by the spatiotemporal coupling coefficient between any two device nodes, and a threshold for the spatiotemporal coupling coefficient is set. Knowledge loading module 13 is used to load cross-disciplinary knowledge graphs, which record multiple historical fault modes and cross-disciplinary anomaly spatiotemporal correlation modes corresponding to each historical fault mode. Anomaly location module 14 is used to locate the anomaly spatiotemporal coordinates in the cross-professional spatiotemporal correlation map when anomalies occur in the multi-professional operation and maintenance data. The associated retrieval module 15 is used to retrieve the cross-professional spatiotemporal association map based on the abnormal spatiotemporal coordinates, according to the preset spatiotemporal range and the spatiotemporal coupling coefficient threshold, and obtain associated operation and maintenance data. The matching module 16 is used to match the associated operation and maintenance data with each of the historical fault modes. The similarity calculation in the matching process includes the contribution of the spatiotemporal coupling coefficient to obtain the matching result. The result output module 17 is used to output the diagnostic results and collaborative treatment plans corresponding to the matching results.

[0070] Specifically, the data loading module 11 is used for: Load multi-disciplinary operation and maintenance data for urban rail transit.

[0071] Specifically, the map construction module 12 is used for: The multi-disciplinary operation and maintenance data includes vehicle operation and maintenance data, power supply operation and maintenance data, civil engineering operation and maintenance data, signaling operation and maintenance data, and electromechanical operation and maintenance data. The spatiotemporal synchronization processing includes standardization processing and spatiotemporal tag synchronization processing. The standardization process includes: converting the original units of each physical quantity in each professional operation and maintenance data into preset standard international units; mapping the original coding system of each professional operation and maintenance data into a unified coding dictionary according to preset standardization mapping rules; unifying the sampling frequency of each professional operation and maintenance data to a preset standard time granularity; and converting the different data formats of each professional operation and maintenance data into a preset unified data structure model. The spatiotemporal tag synchronization processing includes: using the Global Navigation Satellite System as a unified time source to perform synchronous calibration on the data acquisition time of each professional system to obtain a unified time tag; constructing a five-level spatial coding system and converting the original spatial positioning methods of each professional system into the five-level spatial coding system through a preset spatial mapping relationship table to obtain a unified spatial tag; and adding the unified time tag and the unified spatial tag to the standardized data to obtain the spatiotemporal synchronized operation and maintenance data.

[0072] Specifically, the five-level spatial coding system includes line number, station number, up and down direction, relative mileage, and equipment level; For the vehicle professional operation and maintenance data, the train number and kilometer marker are mapped to the five-level spatial coding system through the preset spatial mapping relationship table; For the power supply operation and maintenance data, the substation number and feeder number are mapped to the five-level spatial coding system through the preset spatial mapping relationship table; For the aforementioned engineering maintenance data, the line sections and mileages are mapped to the five-level spatial coding system through the preset spatial mapping relationship table; For the signaling professional operation and maintenance data, the station and signal number are mapped to the five-level spatial coding system through the preset spatial mapping relationship table.

[0073] The spatiotemporal coupling coefficient is determined in the following way: Obtain the time skew tolerance and spatial topology connection distance between any two device nodes; The time dimension association weight is determined based on the time deviation tolerance, and the spatial dimension association weight is determined based on the spatial topology connection distance and the type of equipment along the urban rail transit line. The time-dimensional association weights and the spatial-dimensional association weights are weighted and fused to obtain the spatiotemporal coupling coefficient.

[0074] Furthermore, based on the aforementioned spatiotemporal synchronized operation and maintenance data, a cross-professional spatiotemporal correlation map is constructed, including: The spatiotemporal synchronization operation and maintenance data is stored in association according to the time and space dimensions. The cross-professional spatiotemporal association graph is constructed with devices as nodes and the spatiotemporal coupling coefficient between any two device nodes as edge weights.

[0075] Specifically, the knowledge loading module 13 is used for: Load a cross-disciplinary knowledge graph, which records multiple historical failure modes and the corresponding cross-disciplinary anomaly spatiotemporal correlation patterns for each historical failure mode.

[0076] The anomaly location module 14 is specifically used for: When an anomaly occurs in the multi-disciplinary operation and maintenance data, the anomaly's spatiotemporal coordinates are located in the cross-disciplinary spatiotemporal correlation map.

[0077] The associated retrieval module 15 is specifically used for: Centered on the abnormal spatiotemporal coordinates, the cross-disciplinary spatiotemporal correlation map is retrieved according to the preset spatiotemporal range and the spatiotemporal coupling coefficient threshold to obtain the associated operation and maintenance data, including: A temporal neighborhood is obtained by expanding the abnormal time interval in the abnormal spatiotemporal coordinates according to a preset time radius threshold, and a spatial neighborhood is obtained by expanding the abnormal spatial range in the abnormal spatiotemporal coordinates according to a preset spatial radius threshold. Merge the temporal neighborhood and the spatial neighborhood, and extract the device node pairs in which the spatiotemporal coupling coefficient is greater than or equal to the spatiotemporal coupling coefficient threshold, and determine them as the same spatiotemporal neighborhood; The associated operation and maintenance data is obtained by retrieving all device nodes falling within the same spatiotemporal neighborhood and the corresponding operation and maintenance data of each device node from the cross-professional spatiotemporal correlation map.

[0078] Specifically, the matching module 16 is used for: The associated operation and maintenance data is matched with each of the historical fault modes. The similarity calculation in the matching process includes a contribution term of the spatiotemporal coupling coefficient, including: Extract the temporal sequence and spatial relative position of anomalies in various professional data from the associated operation and maintenance data to form a spatiotemporal association pattern of anomalies to be matched. Obtain the spatiotemporal coupling coefficient between any two device nodes in the associated operation and maintenance data, and calculate the benchmark similarity between the cross-professional spatiotemporal correlation pattern of the abnormal spatiotemporal correlation pattern to be matched and the cross-professional abnormal spatiotemporal correlation pattern of each historical fault pattern. The weighted values ​​of the baseline similarity and the spatiotemporal coupling coefficient are superimposed to obtain the final similarity corresponding to each of the historical fault modes; Extract historical fault patterns whose final similarity is greater than a preset similarity threshold to obtain the matching result.

[0079] Specifically, the result output module 17 is used for: Output the diagnostic results and collaborative treatment plan corresponding to the matching results.

[0080] Specifically, it also includes: When a matching target historical fault pattern exists in the matching results, a fault propagation link diagram is constructed based on the cross-professional anomaly spatiotemporal correlation pattern of the target historical fault pattern. The fault propagation link diagram includes the fault initiation professional identifier, the fault propagation path, the list of affected professional disciplines, and the temporal relationship between each node. When no matching target historical fault mode is found in the matching results, the abnormal spatiotemporal coordinates, the associated operation and maintenance data, and the spatiotemporal sequence of each professional anomaly in the associated operation and maintenance data are combined into a new anomaly mode and added to the cross-knowledge knowledge graph.

[0081] Example 3: This embodiment of the invention also provides a collaborative diagnostic platform for urban rail transit operation and maintenance equipment, which deploys a collaborative diagnostic system for urban rail transit operation and maintenance equipment as described in Example 2.

Claims

1. A collaborative diagnostic method for urban rail transit operation and maintenance equipment, characterized in that, include: Load multi-disciplinary operation and maintenance data for urban rail transit; Spatiotemporal synchronization processing is performed on the multi-disciplinary operation and maintenance data to obtain spatiotemporal synchronized operation and maintenance data, and a cross-disciplinary spatiotemporal correlation graph is constructed based on the spatiotemporal synchronized operation and maintenance data. The edge weights of the cross-disciplinary spatiotemporal correlation graph are determined by the spatiotemporal coupling coefficient between any two device nodes, and a threshold for the spatiotemporal coupling coefficient is set. Load a cross-disciplinary knowledge graph, which records multiple historical failure modes and cross-disciplinary anomaly spatiotemporal correlation patterns corresponding to each historical failure mode; When an anomaly occurs in the multi-disciplinary operation and maintenance data, locate the anomaly's spatiotemporal coordinates in the cross-disciplinary spatiotemporal correlation map; Centered on the abnormal spatiotemporal coordinates, the cross-professional spatiotemporal correlation map is retrieved according to the preset spatiotemporal range and the spatiotemporal coupling coefficient threshold to obtain the associated operation and maintenance data; The associated operation and maintenance data is matched with each of the historical fault modes. The similarity calculation in the matching process includes the contribution of the spatiotemporal coupling coefficient to obtain the matching result. Output the diagnostic results and collaborative treatment plan corresponding to the matching results.

2. The collaborative diagnostic method for urban rail transit operation and maintenance equipment as described in claim 1, characterized in that, The multi-disciplinary operation and maintenance data includes vehicle operation and maintenance data, power supply operation and maintenance data, civil engineering operation and maintenance data, signaling operation and maintenance data, and electromechanical operation and maintenance data. The spatiotemporal synchronization processing includes standardization processing and spatiotemporal tag synchronization processing. The standardization process includes: converting the original units of each physical quantity in each professional operation and maintenance data into preset standard international units; mapping the original coding system of each professional operation and maintenance data into a unified coding dictionary according to preset standardization mapping rules; unifying the sampling frequency of each professional operation and maintenance data to a preset standard time granularity; and converting the different data formats of each professional operation and maintenance data into a preset unified data structure model. The spatiotemporal tag synchronization processing includes: using the Global Navigation Satellite System as a unified time source to perform synchronous calibration on the data acquisition time of each professional system to obtain a unified time tag; constructing a five-level spatial coding system and converting the original spatial positioning methods of each professional system into the five-level spatial coding system through a preset spatial mapping relationship table to obtain a unified spatial tag; and adding the unified time tag and the unified spatial tag to the standardized data to obtain the spatiotemporal synchronized operation and maintenance data.

3. The collaborative diagnostic method for urban rail transit operation and maintenance equipment as described in claim 2, characterized in that, The five-level spatial coding system includes line number, station number, up and down direction, relative mileage, and equipment level; For the vehicle professional operation and maintenance data, the train number and kilometer marker are mapped to the five-level spatial coding system through the preset spatial mapping relationship table; For the power supply operation and maintenance data, the substation number and feeder number are mapped to the five-level spatial coding system through the preset spatial mapping relationship table; For the aforementioned engineering maintenance data, the line sections and mileages are mapped to the five-level spatial coding system through the preset spatial mapping relationship table; For the signaling professional operation and maintenance data, the station and signal number are mapped to the five-level spatial coding system through the preset spatial mapping relationship table.

4. The collaborative diagnostic method for urban rail transit operation and maintenance equipment as described in claim 1, characterized in that, The spatiotemporal coupling coefficient is determined in the following way: Obtain the time skew tolerance and spatial topology connection distance between any two device nodes; The time dimension association weight is determined based on the time deviation tolerance, and the spatial dimension association weight is determined based on the spatial topology connection distance and the type of equipment along the urban rail transit line. The time-dimensional association weights and the spatial-dimensional association weights are weighted and fused to obtain the spatiotemporal coupling coefficient.

5. The collaborative diagnostic method for urban rail transit operation and maintenance equipment as described in claim 1, characterized in that, Based on the aforementioned spatiotemporal synchronized operation and maintenance data, a cross-professional spatiotemporal correlation map is constructed, including: The spatiotemporal synchronization operation and maintenance data is stored in association according to the time and space dimensions. The cross-professional spatiotemporal association graph is constructed with devices as nodes and the spatiotemporal coupling coefficient between any two device nodes as edge weights.

6. The collaborative diagnostic method for urban rail transit operation and maintenance equipment as described in claim 1, characterized in that, Centered on the abnormal spatiotemporal coordinates, the cross-disciplinary spatiotemporal correlation map is retrieved according to the preset spatiotemporal range and the spatiotemporal coupling coefficient threshold to obtain the associated operation and maintenance data, including: A time neighborhood is obtained by expanding the abnormal time interval in the abnormal spatiotemporal coordinates according to a preset time radius threshold, and a spatial neighborhood is obtained by expanding the abnormal spatial range in the abnormal spatiotemporal coordinates according to a preset spatial radius threshold. Merge the temporal neighborhood and the spatial neighborhood, and extract the device node pairs in which the spatiotemporal coupling coefficient is greater than or equal to the spatiotemporal coupling coefficient threshold, and determine them as the same spatiotemporal neighborhood; The associated operation and maintenance data is obtained by retrieving all device nodes falling within the same spatiotemporal neighborhood and the corresponding operation and maintenance data of each device node from the cross-professional spatiotemporal correlation map.

7. The collaborative diagnostic method for urban rail transit operation and maintenance equipment as described in claim 1, characterized in that, The associated operation and maintenance data is matched with each of the historical fault modes. The similarity calculation in the matching process includes a contribution term of the spatiotemporal coupling coefficient, including: Extract the temporal sequence and spatial relative position of anomalies in various professional data from the associated operation and maintenance data to form a spatiotemporal association pattern of anomalies to be matched. Obtain the spatiotemporal coupling coefficient between any two device nodes in the associated operation and maintenance data, and calculate the benchmark similarity between the cross-professional spatiotemporal correlation pattern of the abnormal spatiotemporal correlation pattern to be matched and the cross-professional abnormal spatiotemporal correlation pattern of each of the historical fault patterns. The weighted values ​​of the baseline similarity and the spatiotemporal coupling coefficient are superimposed to obtain the final similarity corresponding to each of the historical fault modes; Extract historical fault patterns whose final similarity is greater than a preset similarity threshold to obtain the matching result.

8. The collaborative diagnostic method for urban rail transit operation and maintenance equipment as described in claim 1, characterized in that, Also includes: When a matching target historical fault pattern exists in the matching results, a fault propagation link diagram is constructed based on the cross-professional anomaly spatiotemporal correlation pattern of the target historical fault pattern. The fault propagation link diagram includes the fault originating professional identifier, the fault propagation path, the list of affected professional disciplines, and the temporal relationship between each node. When no matching target historical fault mode is found in the matching results, the abnormal spatiotemporal coordinates, the associated operation and maintenance data, and the spatiotemporal sequence of each professional anomaly in the associated operation and maintenance data are combined into a new anomaly mode and added to the cross-knowledge knowledge graph.

9. A collaborative diagnostic system for urban rail transit operation and maintenance equipment, characterized in that, A collaborative diagnostic method for urban rail transit operation and maintenance equipment as described in any one of claims 1 to 8, comprising: The data loading module is used to load multi-disciplinary operation and maintenance data for urban rail transit. The graph construction module is used to perform spatiotemporal synchronization processing on the multi-professional operation and maintenance data to obtain spatiotemporal synchronized operation and maintenance data, and to construct a cross-professional spatiotemporal association graph based on the spatiotemporal synchronized operation and maintenance data. The edge weights of the cross-professional spatiotemporal association graph are determined by the spatiotemporal coupling coefficient between any two device nodes, and a threshold for the spatiotemporal coupling coefficient is set. The knowledge loading module is used to load cross-disciplinary knowledge graphs, which record multiple historical fault modes and cross-disciplinary anomaly spatiotemporal correlation patterns corresponding to each historical fault mode. An anomaly localization module is used to locate the anomaly's spatiotemporal coordinates in the cross-professional spatiotemporal correlation map when anomalies occur in the multi-professional operation and maintenance data. The associated retrieval module is used to retrieve the cross-professional spatiotemporal association map based on the abnormal spatiotemporal coordinates, according to a preset spatiotemporal range and the spatiotemporal coupling coefficient threshold, to obtain associated operation and maintenance data. The matching module is used to match the associated operation and maintenance data with each of the historical fault modes. The similarity calculation in the matching process includes the contribution of the spatiotemporal coupling coefficient to obtain the matching result. The result output module is used to output the diagnostic results and collaborative treatment plans corresponding to the matching results.

10. A collaborative diagnostic platform for urban rail transit operation and maintenance equipment, characterized in that, The urban rail transit operation and maintenance equipment collaborative diagnosis platform is equipped with an urban rail transit operation and maintenance equipment collaborative diagnosis system as described in claim 9.

Images