Railway signal equipment whole life cycle data management system based on multi-source data

Abstract

The present application relates to the technical field of railway signal equipment management, and particularly relates to a railway signal equipment whole life cycle data management system based on multi-source data, which comprises a life cycle event fingerprint generation module: accessing multi-source original data, extracting event elements from the multi-source original data, and generating life cycle event fingerprints; a cross-stage causal chain construction module: identifying associated transmission relationships and constructing cross-stage causal chains; a hidden deterioration unit identification module: comparing the cross-stage causal chains with normal causal chains, identifying abnormal chain segments, and determining hidden deterioration units; a life cycle data reorganization module: re-calling data around the hidden deterioration units and reorganizing deterioration tracing data packets; a closed-loop management strategy generation module: generating management strategies corresponding to the railway signal equipment, writing the management strategies and feedback data back to the life cycle event fingerprints, and updating the cross-stage causal chains. The present application improves the intelligent level, risk prediction ability and maintenance decision accuracy.

Description

Technical Field

[0001] This invention relates to the field of railway signaling equipment management technology, and in particular to a railway signaling equipment full lifecycle data management system based on multi-source data. Background Technology

[0002] With the continuous improvement of railway intelligence and digitalization, railway signaling equipment generates a large amount of heterogeneous data during the design, manufacturing, operation, and maintenance stages, including design parameter data, manufacturing test data, operation monitoring data, and maintenance work order data. Current railway signaling equipment management methods mostly adopt an independent management model based on each stage, lacking a unified correlation mechanism between data from different stages. This makes it difficult to continuously track the equipment's status throughout its entire lifecycle and effectively reflect the evolution of hidden degradation. Existing technologies typically rely on single-stage threshold alarms or real-time operating parameter monitoring for identifying anomalies in railway signaling equipment. These methods can only detect explicit anomalies that have reached fault conditions, lacking the ability to effectively identify hidden risks that gradually accumulate over multiple stages due to design deviations, manufacturing errors, and operational fluctuations.

[0003] Meanwhile, significant differences in data structures exist between different business systems, and a lack of a unified causal correlation analysis mechanism across stages makes it difficult to quickly trace the formation path and root cause of equipment anomalies. Existing railway signaling equipment management systems typically lack dynamic feedback and update mechanisms, making it difficult to reuse operational feedback data after maintenance in subsequent risk analysis and model correction. This results in anomaly identification rules remaining fixed for a long time, failing to be continuously optimized as equipment operating status changes. Summary of the Invention

[0004] This invention provides a railway signaling equipment lifecycle data management system based on multi-source data, which can construct a railway signaling equipment lifecycle data management system that can realize the association of railway signaling equipment lifecycle data, identification of hidden degradation, anomaly tracing and closed-loop dynamic updates.

[0005] A railway signaling equipment lifecycle data management system based on multi-source data includes the following modules:

[0006] Lifecycle event fingerprint generation module: It accesses multi-source raw data generated by railway signaling equipment during the design, manufacturing, operation and maintenance phases, extracts event elements such as design constraints, manufacturing and testing deviations, operational fluctuations, fault alarms, maintenance actions and post-maintenance recovery, and merges them according to equipment identity, occurrence time and business stage to generate lifecycle event fingerprints.

[0007] Cross-stage causal chain construction module: Identify the correlation and transmission relationships between design constraints and manufacturing test deviations, manufacturing test deviations and operational fluctuations, operational fluctuations and fault alarms, fault alarms and maintenance actions, and maintenance actions and post-maintenance recovery items, and construct cross-stage causal chains for railway signaling equipment;

[0008] Latent degradation unit identification module: compares the cross-stage causal chain with the normal causal chain, identifies abnormal chain segments that do not meet the fault conditions in a single stage but deviate from the normal causal chain in multiple consecutive stages, and determines the corresponding latent degradation equipment.

[0009] Lifecycle data reorganization module: Retrieves original data from the design, manufacturing, operation and maintenance stages of equipment with hidden degradation, and reorganizes it into degradation traceability data packages according to the transmission order of abnormal chain segments;

[0010] Closed-loop management strategy generation module: Generates management strategies based on deterioration traceability data packets, and writes back the feedback data after the management strategy is executed to the lifecycle event fingerprint for updating the cross-stage causal chain.

[0011] Optionally, the lifecycle event fingerprint generation module specifically includes:

[0012] Heterogeneous data access unit: It is equipped with interface adapters corresponding to the design stage, manufacturing stage, operation stage and maintenance stage, and is used to obtain the multi-source raw data from the product data management system, manufacturing execution system, railway signal centralized monitoring system and maintenance work order management system according to the preset extraction cycle or event triggering method.

[0013] Event Element Analysis Unit: It has a built-in feature mapping rule library corresponding to each business stage, which is used to perform structured semantic analysis on multi-source raw data and extract event elements that can characterize the state changes of railway signaling equipment.

[0014] Fingerprint merging and generation unit: Using the unique identifier of railway signal equipment as the primary key, the parsed event elements are merged according to the chronological order of occurrence and the progressive order of business stages. Duplicate or redundant event elements within the same stage are deduplicated and merged, and stored using an ordered linked list or a graph structure with timestamps to generate the lifecycle event fingerprint.

[0015] Optionally, the data interface adapter in the design phase is connected to the product data management system to obtain design parameters, design constraints, and design redundancy configuration data of the railway signaling equipment; the data interface adapter in the manufacturing phase is connected to the manufacturing execution system to obtain factory test records, process inspection records, and component assembly records; the data interface adapter in the operation phase is connected to the railway signaling centralized monitoring system to obtain operating status parameters, real-time fluctuation data, and fault alarm data; and the data interface adapter in the maintenance phase is connected to the maintenance work order management system to obtain maintenance records, replacement component records, and post-maintenance test results.

[0016] Optionally, the feature mapping rule base includes:

[0017] The data during the design phase is parsed to generate design constraints, which include design voltage, current, temperature range, and safety redundancy requirements.

[0018] Data from the manufacturing stage is analyzed to generate manufacturing test deviation items, which include the absolute and relative deviations of the measured values ​​relative to the factory nominal values.

[0019] Data from the operation phase is parsed to generate operational fluctuation items and fault alarm items. The operational fluctuation items are parameter change sequences that exceed normal operating thresholds, and the fault alarm items are extracted based on a preset fault code mapping table.

[0020] The maintenance phase data is parsed to generate maintenance action items and post-maintenance recovery items. The maintenance action items include the maintenance operation type and the identifier of the replaced parts, and the post-maintenance recovery items include the post-maintenance test values ​​and the stable state of the operating parameters after recovery.

[0021] Optionally, the cross-stage causal chain construction module specifically includes:

[0022] Inter-stage correlation analysis unit: Equipped with a correlation rule engine based on temporal causal inference, used to identify the correlation and transmission relationship between adjacent stages by taking the event elements and timestamps in the life cycle event fingerprint as input;

[0023] Causal chain synthesis unit: All identified transmission edges are merged according to device identity, with design constraint items as the starting node and post-repair recovery items as the ending node. Causal chain storage unit uses a weighted directed graph data structure to store cross-stage causal chains. Each stage causal chain includes the complete stage transmission sequence from design to post-repair recovery and the confidence weight of each transmission edge.

[0024] Causal chain storage unit: The cross-stage causal chain is associated with the corresponding device identity and stored as an instance of a weighted directed graph data structure. An index is established from the node to the original lifecycle event fingerprint, which supports subsequent fast retrieval and comparison by node or path.

[0025] Optionally, the identification of the association transmission relationship between adjacent edge stages specifically includes:

[0026] The design constraints and manufacturing test deviations are aligned in time according to the equipment identity. The correlation coefficient between the design parameter boundary and the measured deviation is calculated. When the correlation exceeds the preset threshold and the deviation direction is consistent with the tolerance direction of the design constraint, a propagation edge from the design constraint to the manufacturing test deviation is generated.

[0027] The manufacturing test deviation item and the operation fluctuation item are matched by sliding match according to the time window. The Granger causality test is used to determine whether the manufacturing stage deviation has a predictive effect on the fluctuation of the operation stage parameters. If so, a transitive edge is generated.

[0028] The operation fluctuation items and fault alarm items are sequence patterned according to the order of event occurrence. When a specific fluctuation pattern frequently accompanies the occurrence of fault alarms within a given time window, a propagation edge from fluctuation to alarm is established.

[0029] The fault alarm items and maintenance action items are semantically matched according to the fault code and the handling measures in the maintenance work order. If the match is successful, a transmission edge from alarm to action is generated.

[0030] The maintenance action item and the post-maintenance recovery item are associated with the maintenance end time and the recovery monitoring start time respectively. It is determined whether the post-maintenance recovery parameters have reached the preset stability standard. If they have, and the replacement or adjustment of the parts in the maintenance action corresponds to the parameter changes in the recovery item, a transfer edge is generated.

[0031] Optionally, the latent degradation unit identification module specifically includes:

[0032] Normal Causal Chain Reference Unit: Stores normal causal chains obtained by training population statistics or machine learning under standard operating conditions for railway signaling equipment of the same type. Each transmission edge in the normal causal chain includes the expected transmission direction, the expected transmission intensity range, and the time delay reference between stages.

[0033] Chain segment deviation detection unit: compares the cross-stage causal chain of the device under test with the normal causal chain in node order, calculates the deviation degree for the transmission edge between each stage, and the deviation degree includes at least the consistency of transmission direction, the difference ratio of transmission strength, and the absolute value of the deviation between the actual time delay and the reference delay; for the independent event elements of each stage, compare whether the design constraint items, manufacturing test deviation items, operation fluctuation items, fault alarm items, maintenance action items, and post-maintenance recovery items exceed the preset single-stage fault judgment threshold of their respective stages, and generates a single-stage limit over-limit flag;

[0034] Multi-stage cumulative deviation determination unit: For inter-stage transmission edges that have not triggered the single-stage limit exceeding flag, the corresponding deviation degree is accumulated according to the device identity and time sequence. When the cumulative deviation score of at least two consecutive stages exceeds the preset cumulative deviation threshold and the independent event elements of each of the at least two consecutive stages do not meet the fault determination conditions, the chain segment corresponding to the at least two consecutive stages is determined as an abnormal chain segment.

[0035] Latent Deterioration Unit Output Unit: The unique identifier of the railway signaling equipment corresponding to the abnormal chain segment is determined as a latent deterioration device, and the start stage, end stage and detailed deviation characteristics of each deviation transmission edge of the corresponding abnormal chain segment are output.

[0036] Optionally, the lifecycle data reorganization module specifically includes:

[0037] Latent Deterioration Unit Parsing Unit: Receives the unique identification of the latent deterioration device and the corresponding abnormal chain segment output by the latent deterioration unit identification module, and parses the original data identifier set corresponding to the start stage, the end stage, and each deviation transmission edge involved in the design constraint item, manufacturing test deviation item, operation fluctuation item, fault alarm item, maintenance action item and post-maintenance recovery item from the abnormal chain segment.

[0038] Cross-stage data recall unit: Based on the unique identifier and the set of original data identifiers, initiate weighted data recall requests to the product data management system in the design stage, the manufacturing execution system in the manufacturing stage, the railway signal centralized monitoring system in the operation stage, and the maintenance work order management system in the maintenance stage, and obtain the corresponding original data records in the order of deviation from the transmission edge in the abnormal chain segment.

[0039] Degradation traceability data packet construction unit: The original data of each stage obtained from the recall are organized according to the transmission order of the abnormal chain segment, with the identity identifier of the hidden degradation device as the root node, each deviation transmission edge as the intermediate node, and the original data record corresponding to each node as the leaf node, to generate a degradation traceability data packet including metadata header.

[0040] Data packet index storage unit: Stores the degraded traceability data packets in a traceability database associated with the device identity, and establishes a reverse index from the abnormal chain segment node to the original lifecycle event fingerprint.

[0041] Optionally, the closed-loop management strategy generation module includes:

[0042] Strategy Decision Unit: Analyzes the abnormal chain segments and original data of each stage in the degradation traceability data packet, and matches and generates management strategies from the strategy rule base according to the deviation characteristics and degree of deviation accumulation of the abnormal chain segments, as well as the equipment type and current operating environment of the hidden degradation equipment.

[0043] Strategy distribution and execution tracking unit: Distributes the generated management strategies to the corresponding execution systems according to their corresponding types. The execution systems specifically include data verification terminals, monitoring parameter configuration servers, maintenance work order management systems, spare parts management systems or decommissioning assessment platforms, and establishes a strategy execution record table to track the execution status, execution time and newly collected feedback data after the execution of each strategy.

[0044] Feedback data write-back unit: After the management strategy is executed, the feedback data generated during and after the execution is structured and encapsulated according to the device identity, occurrence time and business stage, and written back to the life cycle event fingerprint. The feedback data includes at least the changes in device status parameters before and after the strategy execution, the actual effect of maintenance actions, the design constraint deviation correction value confirmed by review, or the actual decommissioning time after decommissioning assessment.

[0045] Causal chain update unit: After detecting the update of the life cycle event fingerprint, it triggers the incremental causal learning process, and re-inputs the newly written feedback data as the new event element into the cross-stage causal chain construction module. It corrects the confidence weight, transmission direction or time delay benchmark of the related transmission edge in the original cross-stage causal chain, and synchronizes the corrected causal chain to the normal causal chain benchmark library unit of the latent degradation unit identification module.

[0046] Optionally, the management strategy includes a data verification strategy, an operation monitoring encryption strategy, a preventive maintenance strategy, a spare parts configuration strategy, and a decommissioning assessment strategy;

[0047] Data verification strategy, used to trigger secondary verification or manual review of the original data at a specified stage;

[0048] Run a monitoring encryption strategy to increase the frequency of monitoring parameter collection or add new monitoring points;

[0049] Preventative maintenance strategies are used to generate maintenance work orders that include maintenance timing, scope, and estimated working hours.

[0050] Spare parts configuration strategy, used to generate a spare parts reserve list and recommended replacement cycle based on deterioration trend prediction;

[0051] The retirement assessment strategy is used to output the remaining service life prediction range and the recommended retirement time window.

[0052] The beneficial effects of this invention are:

[0053] This invention constructs a correlation analysis system of lifecycle event fingerprint, cross-stage causal chain, and latent degradation unit to achieve full lifecycle status correlation tracking of railway signaling equipment during the design, manufacturing, operation, and maintenance stages. It can identify slow-evolving latent degradation problems that are difficult to detect using traditional single-stage monitoring methods, thereby improving the early anomaly identification capability and operational safety of the equipment.

[0054] This invention establishes a normal causal chain benchmark library and combines deviation characteristics such as transmission direction, transmission intensity, and time delay to perform continuous stage cumulative analysis. This enables accurate identification of railway signaling equipment that does not meet the single-stage fault conditions but exhibits a continuous abnormal evolution trend, thereby reducing false alarm rate and missed detection rate and improving the accuracy of hidden risk identification.

[0055] This invention recalls and reassembles the original cross-stage data corresponding to the abnormal chain segment to generate a deterioration traceability data package. Combined with a closed-loop management strategy generation and feedback write-back mechanism, it realizes dynamic closed-loop management of anomaly identification, data traceability, strategy optimization, and causal chain update, thereby improving the intelligence level of railway signaling equipment's full life cycle data management and the accuracy of maintenance decisions. Attached Figure Description

[0056] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only for this invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0057] Figure 1 This is a schematic diagram of the method flow according to an embodiment of the present invention;

[0058] Figure 2 This is a schematic diagram of the system modules in an embodiment of the present invention. Detailed Implementation

[0059] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. For some well-known technologies, those skilled in the art may also use other alternative methods to implement the invention. Moreover, the accompanying drawings are only for more specific description of the embodiments and are not intended to specifically limit the present invention.

[0060] like Figure 1 - Figure 2 As shown, the railway signaling equipment lifecycle data management system based on multi-source data includes the following modules:

[0061] Lifecycle event fingerprint generation module: It accesses multi-source raw data generated by railway signaling equipment during the design, manufacturing, operation and maintenance phases, extracts event elements such as design constraints, manufacturing and testing deviations, operational fluctuations, fault alarms, maintenance actions and post-maintenance recovery, and merges them according to equipment identity, occurrence time and business stage to generate lifecycle event fingerprints.

[0062] The lifecycle event fingerprint generation module specifically includes:

[0063] Heterogeneous data access unit:

[0064] The heterogeneous data access unit is equipped with data interface adapters corresponding to the design, manufacturing, operation, and maintenance phases, respectively, to achieve unified data access between different source systems. Specifically, the data interface adapter for the design phase connects to the product data management system to obtain design parameters, design constraints, and design redundancy configuration data for railway signaling equipment; the data interface adapter for the manufacturing phase connects to the manufacturing execution system to obtain factory test records, process inspection records, and component assembly records; the data interface adapter for the operation phase connects to the railway signaling centralized monitoring system to obtain operating status parameters, real-time fluctuation data, and fault alarm data; and the data interface adapter for the maintenance phase connects to the maintenance work order management system to obtain maintenance records, replacement component records, and post-maintenance test results.

[0065] Multiple sources of raw data are acquired according to a preset extraction period or event triggering method. The periodic extraction method is suitable for periodic operation status data acquisition, while the event triggering method is suitable for data acquisition triggered by fault alarms, maintenance work order generation, or equipment status switching.

[0066] For the incoming data streams, the heterogeneous data access unit generates a unique data time identifier for each piece of raw data. , represented as: ;

[0067] in, Indicates the data collection time; Indicates the identification of railway signaling equipment. Indicates the relevant business stage. Indicates the system identifier of the data source;

[0068] Event Element Analysis Unit:

[0069] The event element analysis unit has a built-in feature mapping rule base corresponding to the design, manufacturing, operation, and maintenance phases. The feature mapping rule base is used to perform unified semantic analysis and standardized mapping on the multi-source raw data generated by railway signaling equipment during the design, manufacturing, operation, and maintenance phases. It mainly includes the following:

[0070] 1. Stage identification rules, used to identify the business stage to which the current raw data belongs, including the design stage, manufacturing stage, operation stage, and maintenance stage;

[0071] 2. Field mapping rules, used to establish the correspondence between the original field names and standard event elements in different business systems;

[0072] 3. Parameter extraction rules are used to extract key parameters such as design voltage, measured deviation value, operating fluctuation parameters, fault codes, maintenance action type and recovery status parameters from the raw data;

[0073] 4. Threshold determination rules are used to identify abnormal features such as operational fluctuations, parameter deviations, and recovery to a stable state;

[0074] 5. Fault semantic mapping rules, used to uniformly map fault codes, fault descriptions, and maintenance measures into standard fault semantic tags;

[0075] 6. Time association rules are used to unify the time format across different business systems and establish the chronological order of events.

[0076] 7. Event element generation rules, used to combine the parsed parameters to generate design constraint items, manufacturing and testing deviation items, operational fluctuation items, fault alarm items, maintenance action items, and post-maintenance recovery items.

[0077] The steps for building a feature mapping rule base include:

[0078] Step 1: Collect historical business data of railway signaling equipment in the product data management system, manufacturing execution system, railway signaling centralized monitoring system, and maintenance work order management system;

[0079] Step two: Standardize the data field names, field types, parameter units, and time formats across different business systems.

[0080] Step 3: Based on the lifecycle management requirements of railway signaling equipment, establish a standard event element model and define the standard fields corresponding to design constraints, manufacturing and testing deviations, operational fluctuations, fault alarms, maintenance actions, and post-maintenance recovery items.

[0081] Step 4: Establish the field mapping relationship between the original fields and standard event elements, and configure the corresponding parameter extraction rules and threshold determination rules;

[0082] Step 5: Based on historical fault cases and maintenance records, establish a semantic mapping relationship between fault codes and maintenance measures;

[0083] Step six involves verifying the results using historical data from normal and abnormal devices, correcting and optimizing the rule matching results, and forming the final feature mapping rule base.

[0084] Table 1 Feature Mapping Rule Base

[0085] Business Phase Original field name Data source system Mapping rules Generate event elements Design phase Rated voltage Product Data Management System Mapped to design voltage parameters Design constraints Design phase Safety redundancy coefficient Product Data Management System Mapped to security redundancy requirements Design constraints Manufacturing stage Factory measured current Manufacturing Execution System The deviation value is generated by comparing it with the nominal value. Manufacturing test deviation items Manufacturing stage Welding inspection results Manufacturing Execution System Extracting deviation levels Manufacturing test deviation items Operational phase Track circuit voltage fluctuation Railway signal centralized monitoring system Exceeding threshold determination generates fluctuation sequence Running fluctuation items Operational phase Fault Codes Railway signal centralized monitoring system Parsing by fault code mapping table Fault alarm items Maintenance phase Replace the relay Maintenance work order management system Extract maintenance action type Maintenance Actions Maintenance phase Voltage test after repair Maintenance work order management system Determine if a stable state has been restored Post-repair restoration items

[0086] For data from the design phase, the event element analysis unit generates design constraint terms based on the digital mapping rules of design parameters. These design constraint terms characterize the operational boundary conditions set for railway signaling equipment during the design phase. Represented as:

[0087] ;

[0088] in, Indicates the design voltage. Indicates the design current. Indicates the design temperature range. This indicates a safety redundancy requirement.

[0089] For data from the manufacturing phase, the event element analysis unit generates manufacturing test deviation items based on the differences between manufacturing test records and factory nominal values, including absolute deviations. Represented as:

[0090] ;

[0091] relative deviation Represented as: ;

[0092] in, This represents the measured value during the manufacturing stage. This indicates the factory-specified value of the corresponding parameter.

[0093] The event element analysis unit uses both absolute deviation and relative deviation as manufacturing test deviation items.

[0094] For data during the operational phase, the event element analysis unit generates operational fluctuation items based on the time series of operational parameters and generates fault alarm items based on the fault degradation code mapping table.

[0095] When the running parameters meet: When the time is right, the corresponding parameter sequence is determined to be a running fluctuation term.

[0096] The fault alarm item maps the original alarm code in the monitoring system to a standardized fault category identifier through the fault code mapping table.

[0097] in, This indicates the current running parameter value. This represents the baseline value for normal operation. This represents the operational fluctuation threshold, used to determine whether the operating parameters of railway signaling equipment deviate from normal operating conditions. The actual value can be set based on the type of railway signaling equipment, parameter stability requirements, and the on-site operating environment. For typical railway signaling equipment such as relays, track circuits, signal drive power supplies, and interlocking control units, the operational fluctuation threshold is typically taken as the corresponding normal operating baseline value. For core control parameters with high safety levels and strict operational stability requirements, the operational fluctuation threshold is set at 5%–15%. For operational parameters significantly affected by ambient temperature, electromagnetic interference, or load changes, the operational fluctuation threshold can be appropriately relaxed to 10%–15%. Given that railway signaling equipment naturally experiences fluctuations under normal operating conditions, it is necessary to avoid misjudging short-term random disturbances as anomalies. Furthermore, objective influencing factors such as temperature changes, power fluctuations, and communication load changes exist in the railway environment, necessitating the maintenance of a reasonable tolerance range. Hidden degradation typically manifests as continuous, small-amplitude deviations. If the threshold is too large, it becomes difficult to identify early degradation trends in a timely manner; if the threshold is too small, it easily generates a large number of false alarms. Using a range of 5%–15% can balance anomaly sensitivity and operational stability.

[0098] For data during the maintenance phase, the event element parsing unit generates maintenance action items and post-maintenance recovery items. The maintenance action items include maintenance operation type, maintenance time, and replacement part identifier; the post-maintenance recovery items include post-maintenance test values ​​and the stable status of operating parameters after recovery.

[0099] The event element analysis unit determines whether to restore a stable state based on parameter fluctuations within the continuous operation time window after maintenance. When the following conditions are met... At that time, it was determined that the railway signaling equipment had entered a stable operating state after restoration.

[0100] The event element parsing unit transmits the parsed design constraints, manufacturing and testing deviations, operational fluctuations, fault alarms, maintenance actions, and post-maintenance recovery items to the fingerprint merging and generation unit.

[0101] in, This represents the standard deviation of the fluctuation of the operating parameters within the time window after recovery. The stability threshold is used to determine whether railway signaling equipment has returned to a stable operating state after maintenance. Its value is set based on the allowable fluctuation range of the operating parameters after maintenance. For railway signaling equipment such as track circuits, power modules, relay interfaces, and voltage acquisition units, the stability threshold is generally set to 1.2 to 2 times the standard deviation of the normal fluctuation of the corresponding operating parameters. For interlocking control equipment with higher safety levels and stricter operational accuracy requirements, the stability threshold is preferably set to 1.2 to 1.5 times the standard deviation of the normal fluctuation. For outdoor signaling equipment significantly affected by the site environment, the stability threshold can be appropriately relaxed to 1.5 to 2 times the standard deviation of the normal fluctuation. Although railway signaling equipment has resumed operation after maintenance, short-term adjustment fluctuations may still exist in the initial stage, so a certain stability buffer zone needs to be maintained. If the stability threshold is set too small, minor fluctuations during the normal recovery process may be misjudged as an unrecovered state; if the stability threshold is set too large, it may mask hidden anomalies that still exist after maintenance. Using 1.2 to 2 times the standard deviation of normal fluctuations as the judgment range can take into account both the accurate judgment of the operational stability after maintenance and the early identification of hidden deterioration risks.

[0102] Fingerprint merging and generation unit:

[0103] Using the unique identifier of railway signaling equipment as the primary key, various event elements output by the event element parsing unit are uniformly merged.

[0104] First, the events are sorted according to their occurrence time. Then, they are organized into stages according to the business progression of the design, manufacturing, operation, and maintenance stages, forming an ordered sequence of events for each stage.

[0105] For event elements that are repeatedly generated within the same stage, the fingerprint merging generation unit uses a feature similarity fusion method to remove duplicates.

[0106] When two event elements are satisfied: When two event elements are identified as duplicate event elements, a fusion process is performed.

[0107] in, This indicates the feature similarity between two event elements. , Each represents a different event element. The event fusion threshold is used to determine whether the similarity of features between two event elements meets the fusion conditions. It is set based on the discreteness of railway signaling equipment lifecycle data, the differences in business stages, and the sensitivity to abnormal events. The event fusion threshold ranges from 0.80 to 0.95. Specifically, for data from the design and manufacturing stages, due to the high degree of standardization of parameter structures and strong field stability, the event fusion threshold is set to 0.90 to 0.95 to avoid incorrect fusion of different design constraints or different manufacturing test deviations. For data from the operation and maintenance stages, due to the significant influence of operational fluctuations, fault alarms, and maintenance actions on on-site conditions, the event fusion threshold can be appropriately reduced to 0.80 to 0.90 to improve the ability to merge recurring abnormal events and continuous maintenance events. When the event fusion threshold is too low, abnormal events occurring at different times and for different reasons are easily misclassified as the same event, weakening the lifecycle event fingerprint's ability to truly reflect the equipment state evolution process. When the event fusion threshold is too high, a large number of recurring events with only slight differences cannot be fused, increasing the redundancy of lifecycle event fingerprint data and reducing the efficiency of subsequent cross-stage causal chain construction. Using a value range of 0.80 to 0.95 can balance event independence, exception retention capability, and lifecycle data compression efficiency.

[0108] The merged event elements are organized into lifecycle event pointers in chronological order. The lifecycle event pointers can be stored using an ordered linked list structure or a graph structure with timestamps, where nodes represent event elements and edges represent time progression relationships and stage transmission relationships.

[0109] Lifecycle event pointers are represented as: ;

[0110] in, This indicates the lifecycle event pointer corresponding to the railway signaling equipment. This represents a set of event elements arranged in chronological order.

[0111] Cross-stage causal chain construction module: Identify the correlation and transmission relationships between design constraints and manufacturing test deviations, manufacturing test deviations and operational fluctuations, operational fluctuations and fault alarms, fault alarms and maintenance actions, and maintenance actions and post-maintenance recovery items, and construct cross-stage causal chains for railway signaling equipment;

[0112] The cross-stage causal chain construction module specifically includes:

[0113] Inter-stage correlation analysis unit:

[0114] The inter-stage correlation analysis unit is equipped with a correlation rule engine based on time-series causal inference, which uses event elements and their timestamps in the lifecycle event fingerprint as input to analyze the correlation and transmission relationships between different business stages.

[0115] First, the design constraints and manufacturing test deviations are aligned chronologically according to the railway signaling equipment identification. Then, the correlation between the corresponding design parameter boundaries and the measured deviations in the manufacturing stage is extracted. The correlation coefficient is used to calculate the correlation between the design parameter boundaries and the manufacturing test deviations, and is expressed as follows:

[0116] ;

[0117] The specific solution involves first obtaining the design constraints and manufacturing test deviations of the same railway signaling equipment during the design and manufacturing phases, and then associating these two types of event elements based on the railway signaling equipment's identification. Next, design parameter boundaries such as design voltage, design current, temperature range, and safety redundancy requirements are extracted from the design constraints, and the measured deviation values ​​of the corresponding parameters are extracted from the manufacturing test deviations. Parameter correspondences are then established according to the same parameter type. After completing the parameter correspondence, each set of design parameter boundaries and corresponding manufacturing test deviations are arranged in chronological order to form a design-manufacturing phase correlation data sequence. Subsequently, each parameter sequence is normalized to eliminate the impact of differences in the dimensions of different parameters on subsequent correlation analysis. Finally, the changing trends of each set of parameters in multiple equipment samples are statistically analyzed to determine whether the measured deviations during the manufacturing phase show a synchronous changing trend with changes in design constraint boundaries.

[0118] If a wider tolerance for a design parameter leads to a greater likelihood of increased measured deviations during the manufacturing stage, or if a lower safety margin for a design parameter makes it easier for deviations to concentrate near the critical range, then a correlation is determined between the design constraint and the manufacturing test deviation. Further analysis is conducted to determine if the direction of the manufacturing test deviation aligns with the tolerance direction in the design constraints. For example, if the design allows for an upper deviation, the manufacturing test results will primarily show positive deviations; or if the design allows for a lower deviation, the manufacturing test results will primarily show negative deviations. When both a high degree of correlation and consistent deviation direction are satisfied between the design parameter boundary and the manufacturing test deviation, a transitive edge is generated from the design constraint to the manufacturing test deviation, and this transitive edge is written into the cross-stage causal chain to characterize the transmission influence of design stage constraints on the formation of manufacturing stage deviations. This represents the correlation coefficient between design constraints and manufacturing / testing deviations. Indicates the first Each design parameter boundary value, This represents the measured deviation value at the corresponding manufacturing stage. This represents the boundary mean of the design parameters. This represents the mean of the measured deviations during the manufacturing stage. This indicates the number of parameters involved in the calculation.

[0119] When the correlation coefficient satisfies: When the direction of manufacturing test deviation is consistent with the tolerance direction in the design constraint, the inter-stage correlation analysis unit generates a transit edge from the design constraint to the manufacturing test deviation.

[0120] in, The threshold for determining the correlation between the design and manufacturing stages is used to judge whether a stable stage-transfer relationship exists between design constraints and manufacturing test deviations. It is set based on the statistical correlation between railway signaling equipment parameters and manufacturing consistency requirements. In practical applications, the threshold for determining the correlation between the design and manufacturing stages is typically between 0.65 and 0.85. For railway signaling equipment with high manufacturing stability requirements, such as relay interface parameters, track circuit electrical parameters, and power module output parameters, the threshold is set to 0.75 to 0.85 to ensure that only design-manufacturing relationships with strong correlations are established as transfer edges. For manufacturing parameters that are significantly affected by process fluctuations, component discreteness, or environmental conditions, the threshold can be appropriately reduced to 0.65 to 0.75 to improve the ability to identify potential latent deviation relationships. If the threshold is set too low, random manufacturing fluctuations may be misjudged as correlations formed by design constraint transfers, resulting in a large number of invalid transfer edges in the cross-stage causal chain. If the threshold is set too high, some early weakly correlated latent degradation features may be missed, reducing the ability to identify subsequent latently degraded equipment. Therefore, using a value range of 0.65 to 0.85 can balance the reliability of the correlation and the ability to retain early anomaly propagation characteristics.

[0121] The inter-stage correlation analysis unit further performs sliding matching of manufacturing test deviation items and operational fluctuation items according to time windows, and uses Granger causality test to determine whether manufacturing stage deviation has a statistically significant predictive effect on operational stage parameter fluctuations.

[0122] The causal relationship between manufacturing test deviations and operational fluctuations is expressed as follows:

[0123] ;

[0124] in, Indicates the time lag order. Represents the random error term. This indicates the fluctuation coefficient at the current moment. Indicates historical operational fluctuation parameters. This represents the historical manufacturing test deviation value. The autoregressive coefficient for operational fluctuations characterizes the degree to which current operational fluctuations of railway signaling equipment inherit from historical operational fluctuations. It is set based on equipment operational inertia, parameter stability, and operational continuity. In practical applications, it is typically between 0.40 and 0.85. For railway signaling equipment with strong operational continuity, such as interlocking control units and track circuit transmitting / receiving modules, the autoregressive coefficient is set between 0.60 and 0.85 to reflect strong historical continuity in operational fluctuations. For outdoor signaling equipment or communication interface equipment significantly affected by external environmental disturbances, the autoregressive coefficient is reduced to between 0.40 and 0.60 to avoid excessive influence of historical fluctuations on the current state. The manufacturing deviation impact coefficient characterizes the degree of influence of manufacturing stage deviations on operational fluctuations in subsequent operational stages, and is set based on the strength of the impact of manufacturing deviations on the long-term stability of the equipment. The manufacturing deviation impact coefficient is set to 0.10 to 0.50. Specifically, for manufacturing issues such as welding deviations, power output deviations, and component assembly deviations that can have a sustained impact on long-term operational stability, the manufacturing deviation impact coefficient is set to 0.30 to 0.50. For manufacturing errors that only have a slight short-term impact, the manufacturing deviation impact coefficient can be set to 0.10 to 0.30. If the autoregressive coefficient of operational fluctuation is set too low, it will be difficult to reflect the continuity of the railway signaling equipment's operating status; if it is set too high, it will easily lead to the model over-reliance on historical fluctuation information. If the manufacturing deviation impact coefficient is set too low, it will not be able to effectively reflect the cumulative impact of implicit deviations in the manufacturing stage on subsequent operating status; if it is set too high, it may amplify occasional manufacturing errors into long-term operational risks. Therefore, the above value range can balance the continuity of operating status, the propagation characteristics of manufacturing deviations, and the accuracy of identifying cross-stage causal relationships.

[0125] When the Granger causality test results meet the statistical significance condition, a transitive edge is generated from the manufacturing test deviation term to the operational fluctuation term. This significance condition refers to determining whether the manufacturing stage deviation can provide a stable and non-random predictive effect on the operational fluctuation after performing a Granger causality test on the manufacturing test deviation and operational fluctuation terms. If the test results show that the prediction error of the operational fluctuation term is significantly reduced after introducing historical data on manufacturing test deviation, and this improvement is not caused by random fluctuations, then a statistically significant relationship is considered to exist between the two. Specifically, the statistical significance condition is usually determined by the significance probability value. When the significance probability value is less than the preset significance level, it indicates that the manufacturing test deviation has a strong predictive influence on subsequent operational fluctuations. At this point, a stage transitive relationship can be identified between the manufacturing stage deviation and the operational fluctuation, and a corresponding transitive edge is generated. Generally, the significance level can be set to 0.05 or 0.01 to ensure a high degree of confidence in the causal relationship and avoid misjudging random operational fluctuations as genuine stage correlations.

[0126] The inter-stage correlation analysis unit then performs sequence pattern mining on the operational fluctuation items and fault alarm items according to the order of event occurrence, in order to identify the correlation between operational fluctuation patterns and fault alarms.

[0127] The frequency of occurrence of a certain operational fluctuation pattern within a time window is expressed as:

[0128] ;

[0129] The inter-stage correlation analysis unit further performs semantic matching between fault alarm items and maintenance action items according to the fault codes and the handling measures in the maintenance work orders.

[0130] When the accompanying frequency exceeds the preset frequency threshold, a transmission edge is established from the operation fluctuation item to the fault alarm item.

[0131] in, This indicates the frequency of accompanying operational fluctuations and fault alarms. This indicates the number of fault alarms that occurred after running the fluctuation mode. This represents the total number of occurrences of the corresponding operational fluctuation pattern, with a frequency threshold of 0.60–0.85. For railway signaling equipment with higher safety levels, the frequency threshold is 0.70–0.85; for railway signaling equipment with significant operational environmental fluctuations, the frequency threshold is 0.60–0.75. A frequency threshold that is too low may misidentify intermittent operational fluctuations as stable fault associations, while a frequency threshold that is too high may miss early accompanying patterns of latent anomalies. Therefore, a range that balances association reliability and anomaly sensitivity is adopted for setting.

[0132] The semantic matching degree between fault alarm items and maintenance action items is represented as follows: ;

[0133] When the accompanying frequency exceeds the preset frequency threshold, a transmission edge is established from the operation fluctuation item to the fault alarm item.

[0134] in, This indicates the semantic matching degree between fault alarm items and maintenance action items. This represents a fault alarm semantic vector. This represents the semantic vector of maintenance actions.

[0135] When the semantic matching degree exceeds the preset matching threshold, a transmission edge is generated from the fault alarm item to the maintenance action item.

[0136] Finally, the inter-stage correlation analysis unit correlates maintenance action items and post-maintenance recovery items according to the maintenance end time and the recovery monitoring start time, and determines whether the post-maintenance recovery parameters have reached the preset stability standard.

[0137] The stability of the restored state after repair is judged by the average fluctuation of the recovery parameters, expressed as:

[0138] ;

[0139] in, This indicates that the average fluctuation of parameters has been restored after maintenance. Indicates the first One recovery monitoring parameter value, This indicates that the baseline value has been restored. This indicates the number of samples collected for monitoring to resume.

[0140] When the recovery parameters reach the preset stability standard, and the replacement or adjustment of the parts in the maintenance action item is consistent with the parameter change object in the post-maintenance recovery item, a transfer edge is generated from the maintenance action item to the post-maintenance recovery item.

[0141] Causal chain synthesis unit:

[0142] All transit edges generated by the inter-stage correlation analysis unit are merged according to the railway signaling equipment identification, and cross-stage causal chains corresponding to the railway signaling equipment are generated; that is, the construction state evolution path is constructed by using the design constraint item as the starting node and the post-maintenance recovery item as the ending node, and adopting a directed acyclic graph structure or path enumeration method.

[0143] Cross-stage causal chain Represented as: ;

[0144] in, Represents the set of event nodes in a causal chain. The set of edges passed during the phase is represented. This represents the set of confidence weights corresponding to each transitive edge.

[0145] For any passing edge, its confidence weight Represented as:

[0146] ;

[0147] The resulting complete stage transmission sequence and corresponding confidence weights are transmitted to the causal chain storage unit.

[0148] in, Indicates a statistical correlation index. This represents the semantic matching degree metric. This indicates the frequency index of the sequence. , , The weighting coefficients for the corresponding indicators are: the weighting coefficients for statistical correlation indicators, semantic matching indicators, and sequence frequency indicators. These coefficients are used to characterize the contribution of different types of association evidence in the cross-stage causal chain. The weighting coefficients for statistical correlation indicators range from 0.40 to 0.60, for semantic matching indicators from 0.20 to 0.35, and for sequence frequency indicators from 0.15 to 0.30, with the sum of the three coefficients preferably being 1. Statistical correlation indicators primarily reflect the numerical correlation strength of parameter changes between different lifecycle stages. Because they directly reflect the objective data transmission relationship between the design, manufacturing, and operation stages, they usually have the highest weight. Semantic matching indicators primarily reflect the consistency of processing logic between fault alarm items and maintenance action items. They are greatly affected by the standardization of text descriptions and work order filling habits, so their weight is moderate. Sequence frequency indicators primarily reflect the recurring patterns between operational fluctuations and fault alarms. They are more suitable for assisting in judging the behavioral pattern association between stages, so their weight is relatively low. If the weight coefficients corresponding to statistical correlation indicators are too low, the dominant role of cross-stage real data association in the causal chain may be weakened; if the weight coefficients corresponding to semantic matching degree indicators are too high, misjudgments may occur due to differences in maintenance record descriptions; if the weight coefficients corresponding to sequence companion frequency indicators are too high, occasional accompanying events may be misidentified as stable causal relationships. Therefore, by adopting the above weight allocation method, the reliability of numerical association, consistency of business semantics, and regularity of operational behavior can be taken into account, thereby improving the accuracy of constructing cross-stage causal chains.

[0149] Causal chain storage unit:

[0150] Cross-stage causal chains are associated and stored with corresponding railway signaling equipment identifiers. Specifically, a weighted directed graph data structure is used to store the cross-stage causal chains, where graph nodes correspond to event elements in the lifecycle event fingerprint, graph edges correspond to stage transmission relationships, and edge weights correspond to the confidence weights of the transmitted edges. Simultaneously, an index mapping relationship is established from cross-stage causal chain nodes to the original lifecycle event fingerprints to support subsequent rapid retrieval, path tracing, and normal causal chain comparison analysis by node, stage path, or abnormal segment chain.

[0151] The node index relationship is represented as follows: ;

[0152] in, This indicates the event node index information. This represents the fingerprint of the corresponding lifecycle event. Indicates the event timestamp. This indicates the stage of the business process.

[0153] Latent degradation unit identification module: compares the cross-stage causal chain with the normal causal chain, identifies abnormal chain segments that do not meet the fault conditions in a single stage but deviate from the normal causal chain in multiple consecutive stages, and determines the corresponding latent degradation equipment.

[0154] The hidden degradation unit identification module specifically includes:

[0155] Normal causal chain benchmark library unit:

[0156] The system stores normal causal chains formed by railway signaling equipment of the same model under standard operating conditions, which serve as a reference for subsequent identification of abnormal chain segments. By performing population statistical analysis or machine learning training on the lifecycle event fingerprints and cross-stage causal chains of a large number of normally operating railway signaling equipment, the system extracts stable and normal transmission relationships between different stages.

[0157] For each stage-transmission edge in a normal causal chain, record the corresponding expected transmission direction, expected transmission strength range, and inter-stage time delay benchmark.

[0158] Normal causal chain Represented as:

[0159] ;

[0160] in, The set of event nodes in a normal causal chain, This represents the set of edges passed during the normal phase. Represents the set of normal transmission strength ranges. The set of reference points for stage time delay is represented; each stage transmission edge in the normal stage transmission edge set includes a transmission start node, a transmission end node, and a corresponding expected transmission direction. The expected transmission direction is used to characterize the state transmission path between different business stages under normal operating conditions.

[0161] The inter-stage time delay benchmark is constructed using the average stage transmission time of historically normal equipment, and is expressed as follows:

[0162] ;

[0163] in, Indicates the time delay benchmark between stages. Indicates the time when the event occurred in the current stage. Indicates the time when the next stage of the event will occur. This indicates the number of normal equipment samples.

[0164] Chain segment deviation detection unit:

[0165] The chain segment deviation detection unit is used to compare the cross-stage causal chain of the railway signaling equipment under test with the normal causal chain segment by segment in order to identify abnormal deviations in the inter-stage transmission relationship.

[0166] The chain segment deviation detection unit first matches the cross-stage causal chain of the device to be detected with the corresponding stage transmission edge in the normal causal chain according to the node order.

[0167] For each stage of the transmission edge, the chain segment deviation detection unit calculates the consistency of the transmission direction, the ratio of transmission strength difference, and the absolute value of the time delay deviation.

[0168] Consistency of transmission direction Represented as: ;

[0169] in, This represents the actual transmission direction vector of the device under test. This represents the expected propagation direction vector in a normal causal chain.

[0170] Transmission intensity difference ratio Represented as: ;

[0171] in, Indicates the actual transmission strength. This indicates the normal transmission strength benchmark.

[0172] The absolute value of the time delay deviation is expressed as: ;

[0173] in, This represents the absolute value of the time delay deviation. This indicates a time delay in the actual stage. This represents the time delay baseline for the normal phase.

[0174] The chain segment deviation detection unit further performs single-stage fault determination based on the independent event elements in each stage.

[0175] Specifically, the design constraints, manufacturing and testing deviations, operational fluctuations, fault alarms, maintenance actions, and post-maintenance recovery items are compared with the preset single-stage fault judgment thresholds for the corresponding stages.

[0176] When the event elements at a certain stage are satisfied: At that time, a single-stage over-limit flag is generated for the corresponding stage.

[0177] in, This indicates the event parameter value at the current stage. This indicates the fault judgment threshold for the corresponding stage, which is set according to the safety tolerance and operational stability requirements of railway signaling equipment at different life stages. The fault judgment threshold for the design stage is 80%–95% of the allowable tolerance range of the design parameters; the fault judgment threshold for the manufacturing stage is 5%–10% of the deviation range of the manufacturer's nominal value; the fault judgment threshold for the operation stage is generally 10%–20% of the normal operating fluctuation range; and the fault judgment threshold for the maintenance stage is 1.5 to 2 times the allowable fluctuation range of the stable parameters after maintenance. Because a threshold that is too low can easily lead to normal fluctuations being misjudged as faults, while a threshold that is too high will miss real anomalies, a setting method close to the safety boundary but retaining a certain buffer margin is adopted to balance fault identification accuracy and system operational stability.

[0178] The chain segment deviation detection unit transmits the calculated deviation characteristics and single-stage over-limit flags to the multi-stage cumulative deviation determination unit.

[0179] Multi-stage cumulative deviation judgment unit:

[0180] The multi-stage cumulative deviation determination unit is used to perform continuous deviation analysis on the inter-stage transmission edge that has not triggered the single-stage over-limit flag, in order to identify hidden deterioration behavior in railway signaling equipment.

[0181] The multi-stage cumulative deviation judgment unit first filters out the stage transmission edge that has triggered the single-stage over-limit flag, and only retains the stage chain segment where none of the independent event elements of each stage have met the fault judgment condition.

[0182] Subsequently, the multi-stage cumulative deviation determination unit performs cumulative calculations on the deviation characteristics in consecutive stages according to the railway signaling equipment identification and the event time sequence.

[0183] Cumulative Deviation Score Represented as: ;

[0184] in, Indicates the first Deviation in transmission direction at each stage Indicates the first The ratio of transmission intensity differences in each stage Indicates the first The absolute value of the time delay deviation at each stage Indicates the number of consecutive stages. Indicates the first The deviation weights for each stage are set progressively higher according to the lifecycle stages: 0.10–0.20 for the design stage, 0.20–0.30 for the manufacturing stage, 0.30–0.40 for the operation stage, and 0.40–0.50 for the maintenance stage. Since the risks of latent degradation in railway signaling equipment are usually more direct in later stages, later stages have a stronger ability to reflect the actual operating status of the equipment and need to be assigned higher weights. At the same time, it is important to avoid excessively amplifying the overall cumulative deviation score due to minor deviations in the early design or manufacturing stages.

[0185] The later the stage, the higher the corresponding deviation weight, in order to enhance the ability to identify hidden anomalies in the operation and maintenance stages.

[0186] When the condition is met: If no stage exceedance flag is triggered in at least two consecutive stages, then the corresponding consecutive stage segment is determined to be an abnormal segment.

[0187] in, The cumulative deviation threshold is set between 1.20 and 2.50. Specifically, for railway signaling equipment with high safety levels and stringent operational stability requirements, the cumulative deviation threshold is set between 1.20 and 1.80; for railway signaling equipment operating in complex environments and allowing for a certain degree of cumulative fluctuation, the cumulative deviation threshold is set between 1.80 and 2.50. Since a threshold that is too low may easily misidentify short-term fluctuations or occasional deviations as latent degradation, while a threshold that is too high may miss latent anomalies that evolve slowly over a long period, a range that simultaneously reflects the cumulative deviation trend over continuous periods and the stability of the equipment is used for setting the threshold.

[0188] Output unit of the hidden degradation unit:

[0189] The railway signaling equipment corresponding to the abnormal chain segment is identified as a latently degraded device, and the corresponding abnormal chain segment information is output. First, the unique identifier of the railway signaling equipment corresponding to the abnormal chain segment is read, and the corresponding railway signaling equipment is identified as a latently degraded device; then, the start stage, end stage, and detailed deviation characteristics of each deviation transmission edge corresponding to the abnormal chain segment are output.

[0190] Among them, deviation characteristics include deviations in transmission direction, differences in transmission intensity, and deviations in stage time delay.

[0191] Exception chain segment Represented as: ;

[0192] in, Indicates the initial stage of the abnormal chain segment. Indicates the termination phase of the abnormal chain segment. This represents the set of deviation propagation edges in the abnormal chain segment.

[0193] Lifecycle data reorganization module: Retrieves original data from the design, manufacturing, operation and maintenance stages of equipment with hidden degradation, and reorganizes it into degradation traceability data packages according to the transmission order of abnormal chain segments;

[0194] The lifecycle data reorganization module specifically includes:

[0195] Hidden degradation unit analysis unit:

[0196] It receives the unique identifier of the hidden degradation device and the corresponding abnormal chain segment output by the hidden degradation unit identification module, and analyzes the deviation transmission relationship in the abnormal chain segment.

[0197] First, read the start stage, end stage, and each deviation propagation edge in the abnormal chain segment, and extract the design constraint items, manufacturing test deviation items, operational fluctuation items, fault alarm items, maintenance action items, and post-maintenance recovery items involved in each deviation propagation edge in the order of the stages.

[0198] For each event element, the identification information of the corresponding original data record is further extracted to form an original data identification set.

[0199] The original data identifier set is represented as: ;

[0200] in, Represents the original data identifier set. It represents a unique data identifier corresponding to the original data record at different stages.

[0201] The exception chain segment is represented as: ;

[0202] in, Indicates an abnormal chain segment. Indicates the initial stage of the abnormal chain segment. This indicates the termination phase of the abnormal chain segment. This represents the set of deviation propagation edges in the abnormal chain segment.

[0203] Cross-stage data recall unit:

[0204] Based on the identification of hidden degradation units and the set of original data indicators, cross-system recall is performed on the original data of railway signaling equipment at different stages of its lifecycle. The cross-stage data recall unit initiates data recall requests to the product data management system (design stage), the manufacturing execution system (manufacturing stage), the centralized railway signaling monitoring system (operation stage), and the maintenance work order management system (maintenance stage). The data recall request includes the equipment identification, original data indicator, and corresponding stage weight information.

[0205] The stage data recall weight table is as follows: ;

[0206] in, This represents the set of data recall weights for each stage. This represents the recall weight during the design phase, with a value ranging from 0.10 to 0.20. This indicates the recall weight during the manufacturing stage, with a value ranging from 0.20 to 0.30. This indicates the recall weight during the operational phase, ranging from 0.30 to 0.40. The weighting for recall during the maintenance phase is 0.30 to 0.45. The sum of the four weights is preferably 1. Design phase data is used to trace the source of initial constraints, so its weight is relatively low; manufacturing phase data is used to analyze the formation process of latent deviations, so its weight is moderate; operation phase data directly reflects the actual deterioration characteristics of the equipment, so its weight is high; maintenance phase data reflects the handling results and recovery status after anomalies, and has a strong role in the closed-loop analysis of latent deterioration, so it is given a high weight.

[0207] The cross-stage data recall unit retrieves the corresponding original data records sequentially according to the deviation propagation edge order in the abnormal chain segment.

[0208] For multiple original data records within the same phase, the cross-phase data retrieval unit sorts them according to their timestamps, resulting in a sorted data sequence. Represented as:

[0209] ;

[0210] in, This represents the data records corresponding to different points in time. This indicates the corresponding timestamp.

[0211] For data records spanning multiple stages, the cross-stage data retrieval unit performs stage alignment based on the time delay benchmark in the abnormal chain segment.

[0212] The stage alignment deviation is expressed as: ;

[0213] in, Indicates the time alignment deviation at each stage. Indicates the actual time interval between stages. This represents the time delay baseline within the abnormal chain segment.

[0214] After completing the cross-stage data retrieval, the cross-stage data retrieval unit transmits the retrieved original data records to the degradation traceability data packet construction unit;

[0215] Degraded traceability data packet construction unit:

[0216] The original data from each stage obtained from the recall are organized according to the transmission order of the abnormal chain segments to construct a degradation tracing data package for hidden degradation tracing analysis.

[0217] First, the data is organized into a tree structure using the identity identifier of the hidden degradation unit as the root node, the deviation propagation edge in the abnormal chain segment as the intermediate node, and the original data record corresponding to each deviation propagation edge as the leaf node.

[0218] Degraded traceable data packets The structure is represented as follows: ;

[0219] in, Represents the root node. Represents the set of intermediate nodes. This represents the set of leaf nodes.

[0220] The root node corresponds to the identity identifier of the hidden degradation unit, the intermediate nodes correspond to the deviation propagation edge in the abnormal chain segment, and the leaf nodes correspond to the original data record.

[0221] Further generate metadata header information; metadata header Represented as: ;

[0222] in, Indicates the equipment model. Indicates the abnormal chain segment identifier. Indicates the timestamp of the data packet generation. Indicates the data version number.

[0223] Data packet index storage unit:

[0224] The degradation tracing data packets are stored in a tracing database associated with the railway signaling equipment identification. The data packet index storage unit first establishes a mapping relationship between the equipment identification and the degradation tracing data packets, and writes the corresponding data packets into the tracing database; then, it establishes a reverse index relationship from the abnormal chain segment node to the original lifecycle event fingerprint.

[0225] Reverse index relationship Represented as:

[0226] ;

[0227] in, Represents a lifecycle event fingerprint. Indicates an abnormal chain segment. This indicates the corresponding degraded traceability data packet.

[0228] Closed-loop management strategy generation module: Generates management strategies based on deterioration traceability data packets, and writes back the feedback data after the management strategy is executed to the lifecycle event fingerprint for updating the cross-stage causal chain.

[0229] The closed-loop management strategy generation module specifically includes:

[0230] Strategic decision-making unit:

[0231] The built-in policy rule engine parses the abnormal chain segments and original data at each stage in the degradation traceability data packet, and matches the corresponding management policy from the policy rule base according to the bias characteristics of the abnormal chain segments, the degree of bias accumulation, the device type of the hidden degradation device, and the current operating environment.

[0232] First, read the bias propagation edge, cumulative bias score, and corresponding stage event elements from the abnormal chain segment, and extract the current operating environment parameters of the railway signaling equipment.

[0233] Equipment Comprehensive Risk Score for: ;

[0234] in, This represents the cumulative bias score corresponding to the anomalous chain segment. Indicates the risk factors of the equipment operating environment. Indicates the degree of abnormality in the current operating status of the equipment. , , The risk weight coefficients are 0.40 to 0.60 for the cumulative deviation score, 0.20 to 0.35 for the operating environment risk factor, and 0.20 to 0.40 for the current abnormality of the equipment's operating status. The sum of the three coefficients is 1. The cumulative deviation score has the highest weight because it can directly reflect the degree of continuous evolution of latent degradation. The operating environment risk factor mainly reflects external influences such as temperature, humidity, electromagnetic interference, and load environment, so its weight is moderate. The current abnormality of the equipment's operating status is used to reflect real-time operating risks, but it is easily affected by short-term fluctuations, so its weight is slightly lower than that of the cumulative deviation score.

[0235] The strategy decision-making unit generates a management strategy by matching the comprehensive risk score of the equipment and the characteristics of the abnormal chain segment from the strategy rule base.

[0236] The data verification strategy triggers secondary verification or manual review of the original data at a specified stage, suitable for situations where design constraints are abnormal or manufacturing / testing deviations pose a risk of data inconsistency. The operational monitoring encryption strategy increases the frequency of monitoring parameter collection or adds monitoring points, suitable for situations where operational fluctuations continuously deviate from the normal causal chain.

[0237] Monitoring and acquisition frequency adjustment coefficient Represented as:

[0238] ;

[0239] in, This indicates the adjusted sampling frequency. This indicates the original acquisition frequency.

[0240] Preventative maintenance strategies are used to generate maintenance work orders, including the timing of maintenance, the scope of maintenance, and the estimated working hours.

[0241] The spare parts configuration strategy is used to generate a spare parts reserve list and recommended replacement cycles based on the deterioration trend prediction results.

[0242] The predicted value for spare parts replacement cycle is expressed as follows:

[0243] ;

[0244] in, This indicates the recommended replacement cycle. Indicates the standard replacement cycle. This indicates the time reduction in lifespan due to latent degradation.

[0245] Remaining life prediction Represented as:

[0246] ;

[0247] in, Indicates the theoretical lifespan of the equipment. This represents the cumulative degradation life loss value.

[0248] The strategy decision-making unit transmits the generated management strategy to the strategy distribution and execution tracking unit.

[0249] Policy distribution and execution tracking unit:

[0250] Different types of management strategies are distributed to the corresponding execution systems, and the strategy execution process is tracked. Data verification strategies are distributed to data verification terminals, operational monitoring encryption strategies are distributed to monitoring parameter configuration servers, preventive maintenance strategies are distributed to maintenance work order management systems, spare parts configuration strategies are distributed to spare parts management systems, and decommissioning assessment strategies are distributed to decommissioning assessment platforms.

[0251] The strategy distribution and execution tracking unit further establishes a strategy execution record table to record and track the entire process of closed-loop management strategies, from generation, distribution, execution to feedback write-back. This table mainly includes the following:

[0252] 1. Policy identification information, including policy number, policy type, and corresponding device identification;

[0253] 2. Strategy source information, including the abnormal chain segment identifier, the corresponding hidden degradation unit identifier, and the trigger time;

[0254] 3. Strategy execution information, including the name of the executing system, execution start time, execution end time, execution status, and information of the personnel responsible for execution;

[0255] 4. Strategy content information, including data verification content, monitoring parameter adjustment content, maintenance scope, spare parts configuration content, and decommissioning assessment results;

[0256] 5. Feedback data collection information, including feedback data identifier, feedback data generation time, equipment status change parameters, and maintenance effect data;

[0257] 6. Closed-loop update information, including whether lifecycle event fingerprint write-back is complete, whether causal chain update is triggered, and update completion time.

[0258] The steps for constructing the strategy execution record table include:

[0259] Step 1: After generating management strategies, the strategy decision-making unit generates a unique strategy identifier for each strategy. Step 2: The strategy distribution and execution tracking unit obtains the corresponding execution system information and establishes an initial record of strategy execution. Step 3: During strategy execution, the strategy execution status, execution time, and execution results are recorded in real time.

[0260] Step 4: After the strategy is executed, collect the corresponding feedback data and establish the association between the feedback data and the strategy identifier;

[0261] Step 5: Write the feedback data write-back status and causal chain update status into the strategy execution record table;

[0262] Step six: Archive and store the completed strategy records for subsequent closed-loop analysis and historical tracing.

[0263] Table 2 Strategy Execution Record Table

[0264] Strategy Number Strategy type Equipment identification Execution status Execution time Feedback results ST-001 Data review strategy SIG-1001 Completed 2026-05-01 The design deviation has been corrected. ST-002 Runtime monitoring encryption policy SIG-1001 Completed 2026-05-01 The sampling frequency has been increased. ST-003 Preventive maintenance strategy SIG-2045 Completed 2026-05-02 The operating status has returned to stability ST-004 Spare parts configuration strategy SIG-3098 In progress 2026-05-03 Spare parts list has been generated ST-005 Retirement Assessment Strategy SIG-4120 Completed 2026-05-04 Retirement time has been updated

[0265] Strategy execution log Represented as:

[0266] ;

[0267] in, Indicates the strategy identifier, Indicates the strategy execution time. Indicates the policy execution status. This indicates the feedback data identifier after execution.

[0268] Policy distribution and execution tracking unit: After the policy is executed, the feedback data is transmitted to the feedback data write-back unit.

[0269] Feedback data write-back unit:

[0270] This is used to rewrite the feedback data generated during and after the execution of management strategies into the lifecycle event fingerprint; firstly, the feedback data is structured and encapsulated according to the railway signal equipment identification, the time of occurrence, and the business stage to which it belongs.

[0271] The feedback data set is represented as follows:

[0272] ;

[0273] in, This represents the set of feedback data. This indicates the changes in device state parameters before and after the strategy execution. This indicates the actual effectiveness data of the maintenance actions. This represents the correction value for design constraint deviations. This indicates the actual retirement time data.

[0274] The feedback data write-back unit then rewrites the feedback data into the lifecycle event fingerprint and generates a new event timestamp and stage identifier.

[0275] Feedback events are represented as:

[0276] ;

[0277] in, Indicates a feedback event, Indicates device identification. Indicates the timestamp of the feedback event. Indicates the relevant business stage. This indicates the content of the feedback data.

[0278] After the feedback data write-back unit completes the write-back, it triggers the causal chain update unit.

[0279] Causal chain update unit:

[0280] After the lifecycle event fingerprint is updated, incremental causal learning and dynamic correction are performed on the original cross-stage causal chain.

[0281] First, newly written feedback events are detected in the lifecycle event fingerprint, and these feedback events are re-input into the cross-stage causal chain construction module as new event elements. Then, the causal chain update unit corrects the transmission edge confidence weight, transmission direction, and time delay benchmark in the original cross-stage causal chain based on the newly added feedback data.

[0282] Pass edge update weights Represented as:

[0283] ;

[0284] in, Represents the original passed edge weights. This indicates the new weight value corresponding to the feedback data. , The updated fusion coefficients are represented by the original passed edge weights, which range from 0.60 to 0.80, and the updated fusion coefficients corresponding to the new weights in the feedback data range from 0.20 to 0.40, with the sum of the two coefficients being 1. The original passed edge weights are derived from cross-stage causal relationships formed by long-term historical data and have high stability, hence they are given higher weights. The new weights corresponding to the feedback data mainly reflect the state changes after the latest strategy is implemented. Although they can reflect real-time evolution trends, they are affected by short-term fluctuations, hence their weights are relatively lower, balancing the stability of the causal chain with dynamic update capabilities.

[0285] The time delay baseline update value is expressed as:

[0286] ;

[0287] in, This represents the updated time delay baseline. Indicates the original time delay reference. This indicates the new time delay value corresponding to the feedback data.

[0288] After the update is completed, the corrected cross-stage causal chain will be synchronized to the normal causal chain benchmark unit in the latent degradation unit identification module for subsequent comparison of abnormal railway signal equipment sections and latent degradation identification.

[0289] Example 1:

[0290] This embodiment selects the turnout idling fault of the ZD6-D type electric switch machine as the management object. This fault mainly manifests as the motor continuously rotating after the control console issues a turnout switching command, the turnout failing to lock into position, the normal and reverse position indicators disappearing, and may be accompanied by a turnout derailment alarm. On-site inspection reveals that this type of fault can be caused by factors such as debris between the switch rail and the stock rail, rusted and jammed rods, insufficient lubrication of the rack block or reducer, excessive rebound force of the switch rail, and slight deformation of the derailment pin. This embodiment uses multi-source data from the design, manufacturing, operation, and maintenance phases to identify continuous deviation behavior where the operating current gradually increases but has not yet reached the single-stage fault condition, and completes cause tracing, repair, and causal chain update after the fault occurs. The following values ​​are used to illustrate the calculation process of this invention and do not constitute a limitation on specific equipment parameters or on-site maintenance standards.

[0291] I. Implementation Targets and Multi-Source Data Access

[0292] A ZD6-D type electric switch machine with equipment number SIG-ZD6D-017 was selected and installed at turnout No. 17 in a certain station. The heterogeneous data access unit was connected to the Product Data Management System (PDM), Manufacturing Execution System (MES), Railway Signal Centralized Monitoring System (CSM), and Maintenance Work Order Management System (MWO) to obtain the equipment's design parameters, factory test data, switching current curves, turnout status, fault alarms, on-site maintenance records, and post-maintenance test data.

[0293] The system generates a unique data identifier for each piece of raw data, represented as:

[0294] ;

[0295] in, Indicates the first Data identifiers for each piece of raw data; Indicates device identification; Indicates the data collection time; Indicates the stage of the lifecycle; This indicates the system identifier from which the data originates.

[0296] For example, an operational current monitoring data point collected on April 10, 2026, is identified as:

[0297] ;

[0298] in, Indicates the operational phase; This refers to the railway signal centralized monitoring system.

[0299] II. Generation of Lifecycle Event Fingerprints

[0300] (a) Generation of design constraint terms

[0301] The event element analysis unit extracts the rated operating voltage, normal operating current reference, upper limit of allowable operating current, friction coupler fault current range, indicator rod notch, and locking test requirements of the electric switch machine from the PDM system.

[0302] The design phase data is shown in Table 1.

[0303] Table 1 Design Constraints for ZD6-D Electric Switch Machine

[0304] parameter Numerical value or requirement Rated operating voltage 160V Normal operating current reference value 1.70A Maximum allowable operating current 2.00A Friction Coupling Fault Current Range 2.30～2.90A Indicator rod notch standard 1.5±0.5mm Locking test requirements 2mm reliable locking, 4mm non-locking

[0305] Design constraints are represented as follows: ;

[0306] Substituting the data into Table 1, we get:

[0307] ;

[0308] in, This indicates a locking constraint of 2mm for reliable locking and 4mm for no locking.

[0309] (ii) Generation of manufacturing test deviation items

[0310] The MES system recorded that the nominal operating current of the device was 1.60A, while the actual measured operating current was 1.72A. Manufacturing stage data is shown in Table 2.

[0311] Table 2 Manufacturing Test Data

[0312] Data Items nominal value Measured value Data source Factory operating current 1.60A 1.72A MES factory test records Extrusion cut pin appearance No deformation No deformation MES Assembly Inspection Record Reducer operating status No abnormal noise No abnormal noise MES Complete Machine Test Record

[0313] The absolute deviation of the generated operating current is:

[0314] ;

[0315] in, Indicates the absolute deviation of the manufacturing operating current; This indicates the measured operating current during the manufacturing stage; This indicates the nominal value of the factory-set operating current.

[0316] Substituting the data, we get:

[0317] ;

[0318] The relative deviation of the generated operating current is:

[0319] ;

[0320] in, This indicates the relative deviation of the manufacturing operating current.

[0321] Substituting the data, we get:

[0322] ;

[0323] The single-stage fault determination threshold during the manufacturing phase is taken as 10% of the manufacturer's nominal value.

[0324] The calculation is as follows:

[0325] ;

[0326] in, This represents the fault determination threshold for the absolute deviation of the operating current during the manufacturing stage.

[0327] because Therefore, the equipment did not meet the single-stage failure conditions during the manufacturing stage.

[0328] However, 7.5% of positive deviations are retained as manufacturing test deviation events.

[0329] ;

[0330] in, Indicates manufacturing test deviation items; This indicates that the single-stage over-limit flag was not triggered during the manufacturing phase.

[0331] (III) Generation of operational fluctuation items during the operation phase

[0332] The railway signal centralized monitoring system collects the conversion current of the equipment according to a preset cycle.

[0333] The operating fluctuation threshold is taken as 10% of the normal operating current reference value, expressed as:

[0334] ;

[0335] in, This indicates the operating current fluctuation threshold.

[0336] An operational fluctuation event is generated when the current operating current satisfies the following equation:

[0337] ;

[0338] in, Indicates time The operating current.

[0339] Some of the monitoring results of this device are shown in Table 3.

[0340] Table 3. Monitoring Results of Operating Current During Operation

[0341] date Operating current / A Relative deviation from the benchmark / A Operational fluctuation determination 2025-01-10 1.72 0.02 normal 2025-02-10 1.78 0.08 normal 2025-03-10 1.88 0.18 Generate running fluctuation items 2025-04-10 1.93 0.23 Generate running fluctuation items 2025-05-10 1.96 0.26 Generate running fluctuation items

[0342] Taking data from May 10, 2025 as an example:

[0343] ;

[0344] Therefore, the running fluctuation term is generated:

[0345] ;

[0346] in, This indicates fluctuations in performance during the operational phase. This indicates the deviation of the operating current from the normal reference value; Indicates the duration of continuous deviation; This indicates that the operating current is showing a continuous upward trend.

[0347] During the operation phase, the single-stage fault threshold is taken as 20% of the normal operating current reference value, and is calculated as follows:

[0348] ;

[0349] in, This indicates that the operating current deviates from the fault threshold during the operation phase. Because... Therefore, the equipment experienced continuous operational deviations, but at that point in time, it had not yet reached the conditions for a single-stage failure.

[0350] (iv) Event merging and lifecycle event fingerprint generation

[0351] For current fluctuation events that are repeatedly uploaded from the same device, in the same direction of operation, and within the same time window, cosine similarity is used to remove duplicates.

[0352] Event similarity is represented as: ;

[0353] in, Indicates an event With the event Feature similarity between them; and Let represent the feature vectors corresponding to the two events. During the operational phase, the event fusion threshold is set to 0.88. When the similarity between two events is 0.93, the following condition is met: Therefore, they are grouped into the same operational fluctuation event, and the maximum operating current, duration, data source, and original data index are retained.

[0354] After merging, the lifecycle event fingerprint of the device is formed:

[0355] ;

[0356] in, This represents the lifecycle event fingerprint of device SIG-ZD6D-017; , and This represents the fluctuations in performance over different time periods; Indicates fault alarm items; Indicates maintenance actions; This indicates items to be restored after repair.

[0357] III. Construction of Cross-Stage Causal Chains

[0358] (a) Correlation between design constraints and manufacturing / testing deviations

[0359] The inter-stage correlation analysis unit selected design current tolerance samples and manufacturing operating current deviation samples from five identical pieces of equipment, as follows:

[0360] ;

[0361] ;

[0362] in, This represents the design current tolerance sample sequence; This represents a sample sequence of manufacturing operating current deviations. The correlation between the design current tolerance and the manufacturing operating current deviation is calculated using the Pearson correlation coefficient.

[0363] ;

[0364] in, This represents the correlation coefficient between the design and manufacturing stages. and These represent the means of the two sample sequences, respectively. Indicates the number of samples.

[0365] ;

[0366] ;

[0367] ;

[0368] ;

[0369] Therefore, we can conclude that:

[0370] ;

[0371] The threshold for determining the design-manufacturing correlation is set to 0.75. Because... Furthermore, the direction of the manufacturing operating current deviation is consistent with the direction of the design current upper tolerance, therefore, the establishment from... point to The passing edge.

[0372] (ii) Correlation between manufacturing and testing deviation items and operational fluctuation items

[0373] The inter-stage correlation analysis unit uses Granger causality test to determine whether the deviation in manufacturing operating current can improve the predictive ability of subsequent operating current fluctuations.

[0374] The running volatility model is represented as:

[0375] ;

[0376] in, Indicates time The operating fluctuation coefficient; Indicates the time lag order; Indicates the first Autoregressive coefficient of the order of operational fluctuations; Indicates the first Manufacturing deviation influence coefficient; Indicates lag Manufacturing test deviations at each stage; This represents the random error term.

[0377] In this embodiment, the lag order is set to 1, and the fitting result is as follows:

[0378] ;

[0379] The sum of squared residuals using only historical data on operational fluctuations is 0.184, while the sum of squared residuals after incorporating manufacturing deviation data is 0.102. The test statistic is expressed as:

[0380] ;

[0381] in, This represents the sum of squared residuals of the constrained model; Represents the sum of squared residuals of the complete model; Indicates the number of newly added explanatory variables; Indicates the number of samples; Indicates the number of parameters in the complete model.

[0382] With a sample size of 24, a new explanatory variable of 1, and a total of 3 model parameters, the following calculations were performed:

[0383] ;

[0384] The test probability is less than 0.01, which is below the significance level of 0.05, indicating that the manufacturing action current deviation has a statistically significant predictive effect on subsequent operational fluctuations. Therefore, a transitive edge is established from the manufacturing test deviation term to the operational fluctuation term.

[0385] (III) Correlation between operational fluctuation items, fault alarm items, and maintenance action items

[0386] Ten instances of continuously rising operating current were found in the historical data of the same type of equipment. Eight of these instances resulted in turnout idling or switching timeout alarms within the subsequent time window.

[0387] The accompanying frequency is expressed as:

[0388] ;

[0389] in, This indicates the frequency at which operational fluctuations and fault alarms occur together; This indicates the number of times an idling or conversion timeout alarm occurred after the operation fluctuated; This indicates the total number of times the corresponding fluctuation pattern occurred. The frequency threshold is set to 0.70, because... Therefore, a transitive edge is established from the operational fluctuation item to the fault alarm item.

[0390] The fault alarm semantic vector and the maintenance action semantic vector are set as follows:

[0391] ;

[0392] ;

[0393] in, The three components respectively represent switching timeout, excessive operating current, and incomplete latching; The three components respectively represent clearing blockage, replenishing lubrication, and retesting lock-up.

[0394] Semantic matching degree is represented as:

[0395] ;

[0396] in, This indicates the semantic matching degree between fault alarm items and maintenance action items.

[0397] ;

[0398] The semantic matching threshold is set to 0.85. Since 0.991 is greater than 0.85, a transitive edge is established from the fault alarm item to the maintenance action item.

[0399] (iv) Calculation of confidence in causal chains

[0400] The overall confidence level of the transitive relationship, considering statistical relevance, semantic matching degree, and sequence association frequency, is expressed as follows:

[0401] ;

[0402] wherein, represents the comprehensive confidence; represents the statistical correlation index; represents the semantic matching degree index; represents the sequence accompanying frequency index; , and respectively represent the weight coefficients of the three indicators. Take , and , the sum of the three weights is 1.

[0403] ;

[0404] thus a cross-stage causal chain among design constraints, manufacturing deviations, operation fluctuations, fault alarms, maintenance actions and post-maintenance recovery is formed as follows:

[0405] ;

[0406] IV. Identification of implicitly deteriorated equipment

[0407] Before the rail switch idling fault occurs formally, the chain segment deviation detection unit compares the cross-stage causal chain of the equipment with the normal causal chain of equipment of the same model segment by segment.

[0408] For the th transmission edge, the consistency of transmission direction is expressed as:

[0409] ;

[0410] wherein, represents the direction consistency of the th transmission edge; represents the actual transmission direction vector; represents the expected transmission direction vector in the normal causal chain. The direction deviation degree is expressed as:

[0411] ;

[0412] wherein, represents the direction deviation degree of the th transmission edge.

[0413] The transmission strength difference ratio is expressed as: ;

[0414] wherein, represents the strength difference ratio of the th transmission edge; represents the actual transmission strength; represents the normal transmission strength benchmark. To eliminate the influence of time dimension, the time delay deviation degree is expressed as:

[0415] ;

[0416] in, Indicates the first Time delay deviation of the transmitted edge; This indicates a time delay in the actual stage; This represents the time delay baseline for the normal phase. The deviation results for each continuous transmission edge are shown in Table 4.

[0417] Table 4. Deviation characteristics of anomalous chain segments

[0418] Passing edges Directional deviation Strength difference ratio Time delay deviation Stage weights Design → Manufacturing 0.15 0.30 0.35 0.20 Manufacturing → Operations 0.22 0.40 0.45 0.30 Operational fluctuations → Fault alarms 0.28 0.50 0.55 0.40 Fault Alarm → Repair 0.20 0.35 0.40 0.50

[0419] The multi-stage cumulative deviation score is expressed as follows:

[0420] ;

[0421] in, This indicates the cumulative deviation score across multiple stages; Indicates the number of consecutive transmitted edges involved in the calculation; Indicates the first The stage weights corresponding to each passing edge. Substituting the data into Table 4, we get:

[0422] ;

[0423] The cumulative deviation threshold is set to 1.20. Since the cumulative deviation score of 1.488 is greater than 1.20, and the single-stage fault threshold was not triggered during the manufacturing and operation phases, the system identifies the equipment SIG-ZD6D-017 as a latently deteriorated device before the turnout idling fault occurs, and outputs the abnormal chain segment from the design phase to the maintenance phase.

[0424] V. Construction of Degraded Tracing Data Packet

[0425] The lifecycle data reassembly module initiates cross-stage data recall to the PDM, MES, CSM, and MWO systems based on the device identification and the original data index in the anomaly chain. The stage data recall weights are represented as follows:

[0426] ;

[0427] in, This represents the set of data recall weights for the design, manufacturing, operation, and maintenance phases, with the sum of the four weights being 1. The design phase is used to trace the source of initial constraints; the manufacturing phase is used to analyze initial deviations; the operation phase is used to reflect the actual degradation process; and the maintenance phase is used to verify the causes of failures and the effectiveness of recovery.

[0428] The degradation traceability data package is organized in a tree structure, as follows: ;

[0429] in, Indicates root node: device SIG-ZD6D-017; Indicate intermediate nodes: deviation propagation edges such as Design-Manufacturing, Manufacturing-Operation, Operation-Alarm; Leaf nodes are represented by: design drawings, factory test records, operating current curves, alarm records, maintenance work orders, and post-maintenance test data; Metadata includes: device model, abnormal chain segment number, generation time, and data version number.

[0430] Data packet analysis shows that the operating current of the equipment already had a positive deviation of 7.5% at the factory. After being put into operation, the operating current gradually increased from 1.72A to 1.96A. The high humidity at the site caused the lubrication of the actuator and rack block to continuously decline, resulting in a deterioration path of increased transmission resistance and inability of the switch rail to lock reliably.

[0431] VI. Closed-loop management strategy generation and on-site handling

[0432] The strategy decision-making unit first normalizes the cumulative deviation score according to the upper limit of the cumulative deviation threshold of 2.50:

[0433] ;

[0434] in, This represents the cumulative deviation score after normalization.

[0435] The degree of abnormality in the current state of the equipment is represented as follows:

[0436] ;

[0437] in, This indicates the degree of abnormality in the current operating status of the equipment.

[0438] The overall risk score for equipment is expressed as follows: ;

[0439] in, This indicates the overall risk score of the equipment; Indicates the risk factors of the equipment operating environment; , and These represent the risk weights for cumulative deviation, operating environment, and the degree of abnormality in the current state, respectively.

[0440] This embodiment takes , , and .

[0441] ;

[0442] The strategy rule base classifies 0.60 to 0.80 as a higher risk level. Since the comprehensive risk score is 0.639, the system generates an operational monitoring encryption strategy and a preventative maintenance strategy: shortening the operating current sampling interval from 60s to 30s; generating preventative maintenance work orders; checking for debris between the switch rail and the main rail, thick edges on the switch rail, and tight contact; checking for jamming of the operating rod, locking rod, indicator rod, and rack block; applying special grease to the reducer, rack block, and rods; checking for deformation of the shear pin; completing 2mm reliable locking and 4mm non-locking tests; and retesting the fault current and fixed / reverse position indicators.

[0443] Three days after the strategy was generated, the equipment experienced a turnout idling incident with an operating current of 2.61A. The motor continued to run, with no indication of either the set or reverse position. On-site personnel first registered the turnout as out of service and confirmed the protective conditions. They then manually checked the jamming location. The inspection revealed corrosion on the operating rod, insufficient lubrication of the rack block, and debris obstructing the switch rail and stock rail. Maintenance personnel removed the debris, worked with track maintenance to repair the burred edge of the switch rail, cleaned and lubricated the rod and rack block, and after confirming that the shear pin was not broken or significantly deformed, the equipment was restored.

[0444] VII. Post-Repair Recovery Assessment

[0445] After the maintenance was completed, the centralized monitoring system continuously collected the operating current five times, and the results are shown in Table 5.

[0446] Table 5. Monitoring results of operating current after maintenance

[0447] Sampling sequence number Operating current / A Absolute deviation relative to 1.70A reference / A 1 1.66 0.04 2 1.69 0.01 3 1.71 0.01 4 1.68 0.02 5 1.70 0.00

[0448] The average absolute fluctuation of the operating current after maintenance relative to the restored reference value is expressed as:

[0449] ;

[0450] in, This indicates the average absolute fluctuation of the operating current after maintenance. Indicates the number of samples collected for monitoring to be restored; Indicates the first Operating current after the second repair. Substituting the data into Table 5, we get:

[0451] ;

[0452] The average operating current after maintenance is:

[0453] ;

[0454] The standard deviation of the operating current fluctuation after maintenance is:

[0455] ;

[0456] in, This represents the standard deviation of the operating current fluctuation after maintenance. The standard deviation under normal conditions is 0.015A, and the steady-state threshold is taken as 1.5 times the standard deviation under normal conditions.

[0457] ;

[0458] in, This indicates the threshold for determining a stable state after maintenance. Because... Furthermore, the operating current is below 2.00A, the 2mm latching test is passed, the 4mm non-latching test is passed, and the fixed / reverse position indicator is normal. Therefore, the post-repair recovery item is generated:

[0459] ;

[0460] in, Indicates the average operating current after maintenance; This indicates that the locking test has passed; The position indicates a fixed or reversed position, indicating normal.

[0461] 8. Feedback Writing and Causal Chain Update

[0462] The feedback data write-back unit encapsulates the results of clearing jamming, replenishing lubrication, restoring operating current, and locking test into feedback events:

[0463] ;

[0464] in, Indicates a feedback event; Indicates device identification; Indicates the repair completion time; Indicates the maintenance phase identifier; This indicates the maintenance actions and the results of the recovery tests.

[0465] The causal chain update unit incrementally updates the weights of the original passing edges based on the feedback data:

[0466] ;

[0467] in, This represents the updated weight of the propagating edge; Indicates the original transitive edge weight; This indicates the new weight value corresponding to the feedback data; and This indicates an update to the fusion coefficients. In this embodiment, the original transitive edge weight is 0.82, and the new weight value is 0.94. and .

[0468] ;

[0469] The original normal time delay benchmark was 24 hours, but the actual feedback delay was 18 hours. The time delay benchmark has been updated as follows:

[0470] ;

[0471] in, This represents the updated time delay baseline; Indicates the original time delay reference; This indicates the actual time delay corresponding to the feedback data.

[0472] ;

[0473] The system synchronizes the updated transmission edge weight of 0.856 and time delay benchmark of 22.2h to the normal causal chain benchmark library for subsequent comparison of abnormal chain segments and identification of hidden degradation in ZD6-D electric switch machines of the same model. Thus, this embodiment completes the closed-loop management of the entire lifecycle from design parameters, manufacturing deviations, operational fluctuations, turnout idling alarms, on-site maintenance, recovery verification to dynamic updates of the causal chain.

[0474] Example 2:

[0475] The following is a specific example of how the railway signaling equipment lifecycle data management system based on multi-source data of this invention is applied to the cable terminal box, a type of railway signaling equipment.

[0476] A certain station uses ZDH-24 type railway signal cable terminal boxes, equipment number THB-2023-0185, which were put into use in 2023. These cable terminal boxes are used for the connection and distribution of track circuits, power cables, and signal control lines, and their operating status directly affects the stability of signal transmission.

[0477] I. Lifecycle Event Fingerprint Generation Process:

[0478] First, the lifecycle event fingerprint generation module obtains the full lifecycle data of the cable terminal box from the Product Data Management System (PDM), Manufacturing Execution System (MES), Railway Signal Centralized Monitoring System (CMS), and Maintenance Work Order Management System (EAM).

[0479] Table 3. Original data for the entire lifecycle of cable terminal boxes

[0480] stage Data Items Data content Design phase Design insulation resistance ≥100MΩ Design phase Upper limit of contact resistance ≤0.05Ω Design phase Operating temperature range -40℃～70℃ Manufacturing stage Measured insulation resistance 108MΩ Manufacturing stage Measured contact resistance 0.047Ω Operational phase Terminal voltage fluctuation ±6.8% Operational phase Contact resistance change rate 18% Operational phase Alarm codes ALM-3205 Maintenance phase Replace terminal block TP-24A Maintenance phase Insulation resistance after repair 115MΩ

[0481] Parsing is performed based on the feature mapping rule base:

[0482] Design constraints are generated by setting upper limits for insulation resistance and contact resistance, and temperature range.

[0483] The difference between the measured value and the nominal value during the manufacturing stage generates a manufacturing test deviation item.

[0484] Voltage fluctuations and changes in contact resistance during the operation phase generate operational fluctuation terms.

[0485] Alarm code ALM-3205 generates a fault alarm item;

[0486] Replacing the terminal block generates a maintenance action item;

[0487] The insulation resistance after repair generates a post-repair recovery item.

[0488] Subsequently, lifecycle event fingerprints were formed by merging them according to device number THB-2023-0185, time sequence, and stage sequence.

[0489] II. The process of constructing a cross-stage causal chain:

[0490] Perform stage correlation analysis on lifecycle event fingerprints.

[0491] Analysis revealed that:

[0492] The upper limit for contact resistance during the design phase is 0.05Ω;

[0493] The measured contact resistance during the manufacturing stage reached 0.047Ω;

[0494] Manufacturing deviations have consistently remained at the edge of design tolerances;

[0495] The rate of change in contact resistance continues to increase during the operation phase.

[0496] The alarm code ALM-3205 then appeared;

[0497] Maintenance records show that the system returned to normal after the terminal block was replaced.

[0498] This allows for the construction of a cross-stage causal chain.

[0499] Table 4. Example of cross-stage causal chain construction results

[0500] starting node Termination Node Relationship Design constraints Manufacturing test deviation items Strong correlation Manufacturing test deviation items Running fluctuation items Granger significant Running fluctuation items Fault alarm items High frequency accompaniment Fault alarm items Maintenance Actions Work order matching Maintenance Actions Post-repair restoration items Recovery successful

[0501] The following state evolution path is formed:

[0502] The contact resistance tolerance was too high during the design phase, the contact resistance was close to the upper limit during the manufacturing phase, the contact resistance continued to increase during the operation phase, the ALM-3205 alarm was triggered, the terminal block was replaced, and the operation returned to normal; thus forming a cross-stage causal chain for this cable terminal box.

[0503] III. Identification process of hidden degradation units:

[0504] In the normal causal chain reference library, the ZDH-24 type cable terminal box of the same model has the following characteristics:

[0505] The contact resistance change rate is less than 10%;

[0506] There was no sustained voltage fluctuation prior to the alarm;

[0507] Manufacturing deviations have a low correlation with operational fluctuations.

[0508] The current performance of the equipment is shown in Table 5:

[0509] Table 5 Comparison of Actual Causal Chains and Normal Causal Chains

[0510] index Normal causal chain Current device Manufacturing deviation correlation 0.42 0.83 Contact resistance change rate 8% 18% Voltage fluctuation amplitude ±2.5% ±6.8% Alarm frequency 1 time / month 7 times / month Stage delay deviation 3 days 17 days

[0511] Calculations revealed that: data during the design phase did not exceed the failure threshold, data during the manufacturing phase did not exceed the failure threshold, and fluctuations during the operation phase did not reach the conditions for a serious failure.

[0512] However, the manufacturing, operation, and alarm stages deviated from the normal causal chain continuously; the cumulative deviation score reached 2.14; the preset cumulative deviation threshold was 1.80.

[0513] Therefore, THB-2023-0185 was identified as a device with hidden degradation;

[0514] The abnormal chain segments are identified as: manufacturing test deviation items, operational fluctuation items, and fault alarm items.

[0515] IV. Lifecycle Data Reassembly Process:

[0516] Data from each stage is retrieved again around the abnormal chain segment.

[0517] The recall includes: design parameters in PDM, factory test records in MES, operation monitoring curves in CMS, and maintenance work order records in EAM.

[0518] Then, the abnormal chain segments are reassembled in sequence: root node: THB-2023-0185, design constraint items, manufacturing and testing deviation items, operational fluctuation items, fault alarm items, maintenance action items, and post-maintenance recovery items; a complete degradation traceability data package is generated.

[0519] Table 6: Example table of metadata headers:

[0520] project content Equipment Model ZDH-24 Equipment Number THB-2023-0185 Exception chain segment number LC-2025-0087 Generation time 2025-06-05 10:25:16 Data version V1.0

[0521] This data package allows maintenance personnel to directly view the complete degradation evolution process from the design phase to the repair and recovery phase.

[0522] V. Closed-loop management strategy generation process:

[0523] Analysis by the strategy decision unit revealed that the equipment exhibits a latent degradation mode characterized by gradually increasing contact resistance, and this mode has been observed multiple times in equipment of the same model.

[0524] Therefore, the following management strategy is generated:

[0525] Data verification strategy: Verify the terminal crimping quality inspection records during the manufacturing stage.

[0526] Operational monitoring encryption strategy: sampling frequency increased from once every 30 minutes to once every 5 minutes; new terminal temperature rise monitoring points added.

[0527] Preventative maintenance strategy: The plan is to complete the terminal fastening inspection of the same batch of terminal boxes within 30 days.

[0528] Spare parts configuration strategy: Increase the inventory of TP-24A terminal blocks by 20%.

[0529] Retirement assessment strategy: The remaining life expectancy is expected to be adjusted from 8 years to about 6 years; it is recommended to arrange an update between 2030 and 2031.

[0530] VI. Three months after the above strategy was implemented, the system collected new feedback data:

[0531] The contact resistance dropped to 0.031Ω, the voltage fluctuation recovered to ±2.1%, the ALM-3205 alarm did not reappear, and 12 units of the same batch of equipment were found to have similar problems.

[0532] Rewrite the above feedback into the lifecycle event fingerprint.

[0533] Subsequently, based on a large amount of historical data, incremental causal learning was initiated, as follows:

[0534] Correcting manufacturing deviations: Propagating edge weights based on operational fluctuations;

[0535] Correcting operational fluctuations: Weighting the edge of fault alarm transmission;

[0536] Update the baseline library for normal causal chains of the same model of equipment.

[0537] After the update, when subsequent cable terminal boxes of the same model show a similar trend of increasing contact resistance, the system can identify the hidden deterioration risk before the fault occurs and generate maintenance strategies in advance, realizing data management of the entire process of railway signal cable terminal boxes from design, manufacturing, operation to maintenance.

[0538] This invention encompasses any substitutions, modifications, equivalent methods, and solutions made within the spirit and scope of this invention. To provide the public with a thorough understanding of this invention, specific details are described in detail in the following preferred embodiments; however, those skilled in the art will fully understand the invention even without these details. Furthermore, to avoid unnecessary misunderstanding of the essence of this invention, well-known methods, processes, procedures, components, and circuits are not described in detail.

[0539] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A railway signaling equipment lifecycle data management system based on multi-source data, characterized in that, Includes the following modules: Lifecycle event fingerprint generation module: It accesses multi-source raw data generated by railway signaling equipment during the design, manufacturing, operation and maintenance phases, extracts event elements such as design constraints, manufacturing and testing deviations, operational fluctuations, fault alarms, maintenance actions and post-maintenance recovery, and merges them according to equipment identity, occurrence time and business stage to generate lifecycle event fingerprints. Cross-stage causal chain construction module: Identify the correlation and transmission relationships between design constraints and manufacturing test deviations, manufacturing test deviations and operational fluctuations, operational fluctuations and fault alarms, fault alarms and maintenance actions, and maintenance actions and post-maintenance recovery items, and construct cross-stage causal chains for railway signaling equipment; Latent degradation unit identification module: compares the cross-stage causal chain with the normal causal chain, identifies abnormal chain segments that do not meet the fault conditions in a single stage but deviate from the normal causal chain in multiple consecutive stages, and determines the corresponding latent degradation equipment. Lifecycle data reorganization module: Retrieves original data from the design, manufacturing, operation and maintenance stages of equipment with hidden degradation, and reorganizes it into degradation traceability data packages according to the transmission order of abnormal chain segments; Closed-loop management strategy generation module: Generates management strategies based on deterioration traceability data packets, and writes back the feedback data after the management strategy is executed to the lifecycle event fingerprint for updating the cross-stage causal chain.

2. The railway signaling equipment full lifecycle data management system based on multi-source data according to claim 1, characterized in that, The lifecycle event fingerprint generation module specifically includes: Heterogeneous data access unit: It is equipped with interface adapters corresponding to the design stage, manufacturing stage, operation stage and maintenance stage, and is used to obtain the multi-source raw data from the product data management system, manufacturing execution system, railway signal centralized monitoring system and maintenance work order management system according to the preset extraction cycle or event triggering method. Event Element Analysis Unit: It has a built-in feature mapping rule library corresponding to each business stage, which is used to perform structured semantic analysis on multi-source raw data and extract event elements that can characterize the state changes of railway signaling equipment. Fingerprint merging and generation unit: Using the unique identifier of railway signal equipment as the primary key, the parsed event elements are merged according to the chronological order of occurrence and the progressive order of business stages. Duplicate or redundant event elements within the same stage are deduplicated and merged, and stored using an ordered linked list or a graph structure with timestamps to generate the lifecycle event fingerprint.

3. The railway signaling equipment full lifecycle data management system based on multi-source data according to claim 2, characterized in that, The data interface adapter in the design phase is connected to the product data management system to obtain the design parameters, design constraints and design redundancy configuration data of the railway signaling equipment. The data interface adapter in the manufacturing stage is connected to the manufacturing execution system to acquire factory test records, process inspection records, and component assembly records; the data interface adapter in the operation stage is connected to the railway signal centralized monitoring system to acquire operating status parameters, real-time fluctuation data, and fault alarm data. The data interface adapter during the maintenance phase is connected to the maintenance work order management system to obtain maintenance records, replacement part records, and post-maintenance test results.

4. The railway signaling equipment full lifecycle data management system based on multi-source data according to claim 2, characterized in that, The feature mapping rule base includes: The data during the design phase is parsed to generate design constraints, which include design voltage, current, temperature range, and safety redundancy requirements. Data from the manufacturing stage is analyzed to generate manufacturing test deviation items, which include the absolute and relative deviations of the measured values ​​relative to the factory nominal values. Data from the operation phase is parsed to generate operational fluctuation items and fault alarm items. The operational fluctuation items are parameter change sequences that exceed normal operating thresholds, and the fault alarm items are extracted based on a preset fault code mapping table. The maintenance phase data is parsed to generate maintenance action items and post-maintenance recovery items. The maintenance action items include the maintenance operation type and the identifier of the replaced parts, and the post-maintenance recovery items include the post-maintenance test values ​​and the stable state of the operating parameters after recovery.

5. The railway signaling equipment full lifecycle data management system based on multi-source data according to claim 1, characterized in that, The cross-stage causal chain construction module specifically includes: Inter-stage correlation analysis unit: Equipped with a correlation rule engine based on temporal causal inference, used to identify the correlation and transmission relationship between adjacent stages by taking the event elements and timestamps in the life cycle event fingerprint as input; Causal chain synthesis unit: All identified transmission edges are merged according to device identity, with design constraint items as the starting node and post-repair recovery items as the ending node. Causal chain storage unit uses a weighted directed graph data structure to store cross-stage causal chains. Each stage causal chain includes the complete stage transmission sequence from design to post-repair recovery and the confidence weight of each transmission edge. Causal chain storage unit: The cross-stage causal chain is associated with the corresponding device identity and stored as an instance of a weighted directed graph data structure. An index is established from the node to the original lifecycle event fingerprint, which supports subsequent fast retrieval and comparison by node or path.

6. The railway signaling equipment full lifecycle data management system based on multi-source data according to claim 5, characterized in that, The specific steps for identifying the correlation and transmission relationship between adjacent edge stages include: The design constraints and manufacturing test deviations are aligned in time according to the equipment identity. The correlation coefficient between the design parameter boundary and the measured deviation is calculated. When the correlation exceeds the preset threshold and the deviation direction is consistent with the tolerance direction of the design constraint, a propagation edge from the design constraint to the manufacturing test deviation is generated. The manufacturing test deviation item and the operation fluctuation item are matched by sliding match according to the time window. The Granger causality test is used to determine whether the manufacturing stage deviation has a predictive effect on the fluctuation of the operation stage parameters. If so, a transitive edge is generated. The operation fluctuation items and fault alarm items are sequence patterned according to the order of event occurrence. When a specific fluctuation pattern frequently accompanies the occurrence of fault alarms within a given time window, a propagation edge from fluctuation to alarm is established. The fault alarm items and maintenance action items are semantically matched according to the fault code and the handling measures in the maintenance work order. If the match is successful, a transmission edge from alarm to action is generated. The maintenance action item and the post-maintenance recovery item are associated with the maintenance end time and the recovery monitoring start time respectively. It is determined whether the post-maintenance recovery parameters have reached the preset stability standard. If they have, and the replacement or adjustment of the parts in the maintenance action corresponds to the parameter changes in the recovery item, a transfer edge is generated.

7. The railway signaling equipment full lifecycle data management system based on multi-source data according to claim 1, characterized in that, The hidden degradation unit identification module specifically includes: Normal Causal Chain Reference Unit: Stores normal causal chains obtained by training population statistics or machine learning under standard operating conditions for railway signaling equipment of the same type. Each transmission edge in the normal causal chain includes the expected transmission direction, the expected transmission intensity range, and the time delay reference between stages. Chain segment deviation detection unit: compares the cross-stage causal chain of the device under test with the normal causal chain in node order, calculates the deviation degree for the transmission edge between each stage, and the deviation degree includes at least the consistency of transmission direction, the difference ratio of transmission strength, and the absolute value of the deviation between the actual time delay and the reference delay; for the independent event elements of each stage, compare whether the design constraint items, manufacturing test deviation items, operation fluctuation items, fault alarm items, maintenance action items, and post-maintenance recovery items exceed the preset single-stage fault judgment threshold of their respective stages, and generates a single-stage limit over-limit flag; Multi-stage cumulative deviation determination unit: For inter-stage transmission edges that have not triggered the single-stage limit exceeding flag, the corresponding deviation degree is accumulated according to the device identity and time sequence. When the cumulative deviation score of at least two consecutive stages exceeds the preset cumulative deviation threshold and the independent event elements of each of the at least two consecutive stages do not meet the fault determination conditions, the chain segment corresponding to the at least two consecutive stages is determined as an abnormal chain segment. Latent Deterioration Unit Output Unit: The unique identifier of the railway signaling equipment corresponding to the abnormal chain segment is determined as a latent deterioration device, and the start stage, end stage and detailed deviation characteristics of each deviation transmission edge of the corresponding abnormal chain segment are output.

8. The railway signaling equipment full lifecycle data management system based on multi-source data according to claim 1, characterized in that, The lifecycle data reorganization module specifically includes: Latent Deterioration Unit Parsing Unit: Receives the unique identification of the latent deterioration device and the corresponding abnormal chain segment output by the latent deterioration unit identification module, and parses the original data identifier set corresponding to the start stage, the end stage, and each deviation transmission edge involved in the design constraint item, manufacturing test deviation item, operation fluctuation item, fault alarm item, maintenance action item and post-maintenance recovery item from the abnormal chain segment. Cross-stage data recall unit: Based on the unique identifier and the set of original data identifiers, initiate weighted data recall requests to the product data management system in the design stage, the manufacturing execution system in the manufacturing stage, the railway signal centralized monitoring system in the operation stage, and the maintenance work order management system in the maintenance stage, and obtain the corresponding original data records in the order of deviation from the transmission edge in the abnormal chain segment. Degradation traceability data packet construction unit: The original data of each stage obtained from the recall are organized according to the transmission order of the abnormal chain segment, with the identity identifier of the hidden degradation device as the root node, each deviation transmission edge as the intermediate node, and the original data record corresponding to each node as the leaf node, to generate a degradation traceability data packet including metadata header. Data packet index storage unit: Stores the degraded traceability data packets in a traceability database associated with the device identity, and establishes a reverse index from the abnormal chain segment node to the original lifecycle event fingerprint.

9. The railway signaling equipment full lifecycle data management system based on multi-source data according to claim 1, characterized in that, The closed-loop management strategy generation module includes: Strategy Decision Unit: Analyzes the abnormal chain segments and original data of each stage in the degradation traceability data packet, and matches and generates management strategies from the strategy rule base according to the deviation characteristics and degree of deviation accumulation of the abnormal chain segments, as well as the equipment type and current operating environment of the hidden degradation equipment. Strategy distribution and execution tracking unit: Distributes the generated management strategies to the corresponding execution systems according to their corresponding types. The execution systems specifically include data verification terminals, monitoring parameter configuration servers, maintenance work order management systems, spare parts management systems or decommissioning assessment platforms, and establishes a strategy execution record table to track the execution status, execution time and newly collected feedback data after the execution of each strategy. Feedback data write-back unit: After the management strategy is executed, the feedback data generated during and after the execution is structured and encapsulated according to the device identity, occurrence time and business stage, and written back to the life cycle event fingerprint. The feedback data includes at least the changes in device status parameters before and after the strategy execution, the actual effect of maintenance actions, the design constraint deviation correction value confirmed by review, or the actual decommissioning time after decommissioning assessment. Causal chain update unit: After detecting the update of the life cycle event fingerprint, it triggers the incremental causal learning process, and re-inputs the newly written feedback data as the new event element into the cross-stage causal chain construction module. It corrects the confidence weight, transmission direction or time delay benchmark of the related transmission edge in the original cross-stage causal chain, and synchronizes the corrected causal chain to the normal causal chain benchmark library unit of the latent degradation unit identification module.

10. The railway signaling equipment full lifecycle data management system based on multi-source data according to claim 9, characterized in that, The management strategies include data verification strategies, operation monitoring encryption strategies, preventive maintenance strategies, spare parts configuration strategies, and decommissioning assessment strategies. Data verification strategy, used to trigger secondary verification or manual review of the original data at a specified stage; Run a monitoring encryption strategy to increase the frequency of monitoring parameter collection or add new monitoring points; Preventative maintenance strategies are used to generate maintenance work orders that include maintenance timing, scope, and estimated working hours. Spare parts configuration strategy, used to generate a spare parts reserve list and recommended replacement cycle based on deterioration trend prediction; The retirement assessment strategy is used to output the remaining service life prediction range and the recommended retirement time window.

Images