A BIM-based railway power supply equipment operation and maintenance platform

Abstract

The present application relates to the technical field of railway operation and maintenance, and discloses a railway power supply equipment operation and maintenance platform based on BIM, which comprises a power supply BIM model construction module, a power supply unit data acquisition and fusion module, a power supply health state evaluation module, a fault positioning and tracing module and a visual interactive dispatching module. The present application directly pushes the three-dimensional coordinates of the faulty equipment to the terminal of the operation and maintenance personnel by setting the power supply BIM model construction module and the visual interactive dispatching module. The present application constructs a fusion time sequence state feature matrix containing power transformation equipment, feeder cable and overhead contact system equipment by setting the power supply unit data acquisition and fusion module and the fault positioning and tracing module, thereby solving the problem of fault positioning deviation caused by the lack of inter-device conduction correlation analysis in the single-device monitoring mode. The present application introduces a predicted pantograph state index to adaptively adjust the comprehensive health degree evaluation threshold, thereby solving the problem of disconnection between the fixed evaluation standard and the operation condition.

Description

Technical Field

[0001] This invention relates to the field of railway operation and maintenance technology, and in particular to a BIM-based railway power supply equipment operation and maintenance platform. Background Technology

[0002] The railway traction power supply system is the core energy security facility for electrified railways, consisting of three main parts: traction substations, feeder cables, and the overhead contact system. The traction substations step down high-voltage electricity, which is then transmitted to the overhead contact system via feeder cables. The pantograph then connects the overhead contact system to the electric locomotive, achieving dynamic electrical connection and completing the final power transmission. The operational status of the railway power supply equipment directly affects the safety and efficiency of railway transportation.

[0003] Currently, the operation and maintenance management of railway power supply equipment mainly adopts two methods: manual inspection and sensor-based single-device monitoring. Manual inspection involves maintenance personnel checking the power supply equipment point-by-point along the railway line at fixed intervals, using methods such as visual observation, sound identification, temperature measurement by touch, and odor detection to determine if there are any abnormalities. Once an abnormality is found, the fault information must be reported level by level, and the dispatching department will notify the corresponding work area personnel to handle the issue. Sensor-based single-device monitoring involves deploying sensors on the power supply equipment, uploading the collected operating status data to an industrial cloud platform, and using threshold alarms or neural network models to independently analyze and diagnose faults in individual devices. However, both of these methods have significant technical drawbacks: First, regarding manual inspections, the inspection cycle is fixed and the intervals are long, making it impossible to continuously monitor the status of power supply equipment. Moreover, the inspection results depend on the personal experience of maintenance personnel. At the same time, after an anomaly is discovered, it is necessary to go through multiple links such as reporting, scheduling, and notification before the maintenance task can be passed on to the specific personnel to perform it. There is a large time delay between the discovery of the fault and the maintenance response, resulting in extremely low maintenance efficiency. Secondly, regarding single-device monitoring, because sensor data from each device is collected, stored, and analyzed independently, when sensor data from a particular device (such as downstream overhead contact line equipment) shows an anomaly, the system directly determines that the device has malfunctioned. It cannot distinguish whether the anomaly is caused by the device's own degradation or by conductive interference resulting from a fault in upstream substation equipment or feeder cables propagating along the electrical conduction path to the measuring point. Due to the lack of correlation analysis capabilities between upstream and downstream devices in the same power transmission link, fault diagnosis results suffer from severe locational biases. Maintenance personnel frequently rush to the alarm point only to find that the device is not faulty, while the real fault source lies in other upstream equipment, resulting in a significant waste of ineffective inspection and maintenance resources. Third, existing technologies use fixed and uniform evaluation standards when assessing the health status of power supply equipment, without considering the impact of differences in the actual working condition of the train pantograph on the evaluation results. In actual operation, the wear degree of the pantograph carbon sliding plate, the pantograph-catenary contact pressure, and the pantograph raising action time vary significantly depending on the train's operating conditions. The worse the pantograph condition (e.g., severe wear of the carbon sliding plate), the higher the requirements for the power supply quality of the catenary (e.g., pantograph-catenary contact stability) to ensure reliable current collection. However, existing technologies use a uniform health evaluation standard regardless of the pantograph condition. When the pantograph condition is poor, the evaluation standard is not raised accordingly, which may lead to a mismatch between the actual power supply quality of the power supply unit and the actual current collection quality of the pantograph, which may not be identified and pose a potential hazard to train safety. When the pantograph condition is good, the evaluation standard is not lowered accordingly, which may lead to power supply units that actually meet the power supply requirements being over-evaluated as unhealthy, triggering unnecessary maintenance procedures. Because existing technologies lack an adaptive linkage mechanism between pantograph status and evaluation standards, discrepancies between assessment results and actual operating conditions persist. It is impossible to dynamically match the corresponding power quality evaluation standards according to the actual power demand of the pantograph, resulting in the inability to guarantee the accuracy of health status assessment and the pertinence of operation and maintenance decisions. Summary of the Invention

[0004] This invention provides a BIM-based railway power supply equipment operation and maintenance platform to solve at least one of the aforementioned technical problems.

[0005] To address the aforementioned technical problems, this invention provides a BIM-based railway power supply equipment operation and maintenance platform, comprising: The power supply BIM model building module is used to build power supply BIM models of each power supply unit along the railway line based on the standardized railway power supply BIM component family library, 3D spatial drawings of railway power supply equipment and electrical connection diagrams of railway power supply equipment. The power supply unit data acquisition and fusion module is used to collect time-series data of various types of substation equipment, various types of feeder cables and various types of contact network equipment of each power supply unit in each monitoring period, and fuse them to obtain the fused time-series status feature matrix of each power supply unit in each monitoring period. The power supply health status assessment module is used to obtain the comprehensive health score of each power supply unit in each monitoring period based on the fused temporal state feature matrix of each power supply unit and the trained comprehensive health score model. By adaptively adjusting the comprehensive health evaluation threshold of the current monitoring period through the predicted pantograph state index when the train runs to the current power supply unit to be evaluated in each monitoring period, the module obtains the adaptive comprehensive health evaluation threshold for each monitoring period. Based on the comprehensive health score of each power supply unit in each monitoring period and the adaptive comprehensive health evaluation threshold, the module outputs the power supply health status assessment result of each power supply unit in each monitoring period. The fault location and tracing module is used to trace and locate faults in equipment based on the fusion time-series state feature matrix of each power supply unit and the corresponding power supply BIM model for each monitoring period, to determine the defect level of the faulty equipment and to provide the optimal maintenance strategy. The visual interactive scheduling module is used to visually identify and locate faulty equipment in the power supply BIM model based on the fault location and tracing results, retrieve the inventory location of the corresponding spare parts for the faulty equipment, match the best maintenance personnel, generate maintenance work orders and push them to the maintenance personnel's terminals.

[0006] Preferably, the power supply BIM model building module includes: The component family library acquisition submodule is used to acquire standard 3D model components of each power supply unit, each substation equipment, each feeder cable and its accessories, and each contact network equipment in each power supply unit along the railway line, based on the standardized railway power supply BIM component family library. Each standard 3D model component contains the corresponding equipment type and the corresponding equipment model. The 3D spatial coordinate acquisition submodule is used to acquire the actual 3D spatial coordinates of each power supply unit, each feeder cable and its accessories, and each contact network equipment within the railway line based on the 3D spatial drawings of the railway power supply equipment, as well as the spatial location coordinates and corresponding quantities of spare parts for each power supply unit, each feeder cable and its accessories, and each contact network equipment. The electrical connection relationship acquisition submodule is used to acquire the electrical connection relationships between electrical components and equipment in each power supply unit along the railway line based on the electrical connection diagram of the railway power supply equipment. The BIM model assembly generation submodule is used to take each standard 3D model component as the basic component unit, place each basic component unit in the corresponding spatial position according to the 3D spatial coordinates of each electrical component and equipment, establish directed connection edges between each basic component unit according to the electrical connection relationship between each electrical component and equipment, and generate the power supply BIM model of each power supply unit along the railway line.

[0007] Preferably, the power supply unit data acquisition and fusion module includes: The substation data acquisition submodule is used to collect data on the traction transformer oil temperature, dissolved acetylene concentration in the traction transformer oil, dissolved hydrogen concentration in the traction transformer oil, grounding current of the traction transformer core, partial discharge quantity inside the GIS, vibration spectrum of the GIS cabinet, SF6 gas concentration, and leakage current of the surge arrester in each power supply unit at each monitoring time during each monitoring period. It also arranges the same type of sensor data at all monitoring times within the same monitoring period in chronological order to form the time sequence data of each type of substation equipment in each monitoring period, and arranges the sensor data of all types of substation equipment at the same monitoring time within the same monitoring period in the order of the preset sensor channels to form the time sequence vector of the substation equipment at the corresponding monitoring time of the corresponding monitoring period. The feeder data acquisition submodule is used to collect sensor data on the distributed optical fiber temperature of the feeder cable sheath, the feeder cable grounding leakage current, and the feeder cable insulation loss coefficient of the feeder cable in each power supply unit at each monitoring time during each monitoring period. It also arranges the same type of sensor data at all monitoring times within the same monitoring period in chronological order to form the time sequence data of each type of feeder cable for each monitoring period. Finally, it arranges the sensor data of all types of feeder cables at the same monitoring time within the same monitoring period in the order of the preset sensor channels to form the feeder cable time sequence vector for the corresponding monitoring time of the corresponding monitoring period. The overhead contact line data acquisition submodule is used to collect data on the overhead contact line equipment in each power supply unit at each monitoring time during each monitoring period, including the tension of the overhead contact line droppers, the vibration amplitude of the overhead contact line brackets, the contact pressure of the overhead contact line pantograph, the leakage current of the overhead contact line insulators, the wear thickness of the overhead contact line conductors, and the tension of the overhead contact line catenary. It also arranges the same type of sensor data at all monitoring times within the same monitoring period in chronological order to form the chronological data of each type of overhead contact line equipment in each monitoring period. Finally, it arranges the sensor data of all types of overhead contact line equipment at the same monitoring time within the same monitoring period in the order of the preset sensor channels to form the chronological vector of the overhead contact line equipment at the corresponding monitoring time of the corresponding monitoring period. The time-series feature fusion submodule is used to concatenate the time-series vectors of substation equipment, feeder cable, and contact network equipment of the same power supply unit at the same monitoring time in the same monitoring period and at the same monitoring moment in the feature dimension to generate a fused time-series state feature row vector for each monitoring moment. The fused time-series state feature row vectors of all monitoring moments in the monitoring period are stacked in time sequence to obtain the fused time-series state feature matrix of each power supply unit in each monitoring period.

[0008] Preferably, the power supply health status assessment module includes: The health score acquisition submodule is used to input the fused time-series status feature matrix of each power supply unit in each monitoring period into the trained comprehensive health score model, and output the comprehensive health score of each power supply unit through the trained comprehensive health score model. The pantograph status evaluation submodule is used to collect various types of operating status data of the train pantograph during each monitoring period, and calculate the current pantograph status index of the train pantograph during each monitoring period based on the remaining thickness of the pantograph carbon sliding plate, the pantograph-catenary contact pressure and the pantograph lifting action time in the various types of operating status data. The adaptive evaluation threshold determination submodule is used to calculate the comprehensive health evaluation threshold for each monitoring period based on the pantograph status index and the basic comprehensive health evaluation threshold of the train pantograph for each monitoring period. The status assessment result judgment submodule is used to calculate the difference between the comprehensive health score of each power supply unit and the adaptive comprehensive health evaluation threshold for each monitoring period. Based on the preset difference range into which the difference falls, the power supply health status assessment level corresponding to each power supply unit for each monitoring period is output.

[0009] Preferably, the adaptive evaluation threshold determination submodule includes: The train status data acquisition unit is used to obtain the average speed of the train during its operation, which is used as the average speed of the train. It also obtains the distance from the current position of the train's pantograph to the corresponding power supply unit at the end of each monitoring period, which is used as the distance the train needs to travel in each monitoring period. The state index prediction execution unit is used to input the train's average driving speed, the train's distance to be traveled in each monitoring period, and the current pantograph state index of the train's pantograph in each monitoring period into the trained pantograph state index prediction model to obtain the predicted pantograph state index when the train runs to the current power supply unit to be evaluated. The adaptive adjustment coefficient calculation unit is used to calculate the adaptive adjustment coefficient for the corresponding monitoring period based on the predicted pantograph state index when the train runs to the current power supply unit to be evaluated. The adaptive evaluation threshold calculation unit is used to calculate the comprehensive health evaluation threshold for each monitoring period based on the basic comprehensive health evaluation threshold and the adaptive adjustment coefficient of the corresponding monitoring period.

[0010] Preferably, the fault location and tracing module includes: The fault screening and judgment submodule is used to receive the power supply health status assessment results of each power supply unit in each monitoring period output by the power supply health status assessment module, and to screen out unhealthy power supply units whose comprehensive health score is lower than the adaptive comprehensive health evaluation threshold, mark them as power supply units to be repaired, and trigger the fault tracing process. The fault equipment tracing submodule is used to compare the standard fusion time-series status feature matrix of each power supply unit in a healthy state with the fusion time-series status feature matrix of the corresponding power supply unit in the current monitoring period, determine the position of the sliding window with the largest deviation from the standard fusion time-series status feature matrix in the current monitoring period, and back-map the sliding window position to the corresponding equipment type and sensing channel to determine the equipment number and equipment type of the faulty equipment. The fault equipment location submodule is used to query the three-dimensional spatial coordinates and power supply unit information of the fault equipment in the power supply BIM model based on the equipment number of the fault equipment, and determine the operation and maintenance work area corresponding to the fault equipment. The defect level determination submodule is used to calculate the defect level value of the faulty equipment based on the similarity value of the sliding window corresponding to the faulty equipment, and to determine the defect level of the faulty equipment based on the preset defect level range into which the defect level value falls. The maintenance strategy determination submodule is used to match the optimal maintenance strategy from a preset maintenance strategy knowledge base based on the equipment type and defect level of the faulty equipment.

[0011] Preferably, the fault equipment tracing submodule includes: The traversal unit determination unit is used to determine the size of the traversal unit of the sliding window based on the evaluation level of the power supply unit corresponding to the comprehensive health status label of the power supply unit in the power supply health status evaluation results. The traversal unit is a rectangular window, and its row height and column width are determined according to the evaluation level. The similarity traversal execution unit is used to synchronously traverse the fusion time-series state feature matrix and the corresponding standard fusion time-series state feature matrix of the power supply unit to be inspected during the current monitoring period, using the rectangular sliding window determined by the traversal unit as the traversal unit. Each traversal extracts the current sub-matrix of the sliding window coverage area in the fusion time-series state feature matrix and the standard sub-matrix at the same position in the standard fusion time-series state feature matrix, calculates the similarity between the current sub-matrix and the standard sub-matrix, and records the similarity value corresponding to each sliding window position. The source identification and judgment unit is used to compare the similarity values ​​of all sliding window positions, locate the sliding window position with the smallest similarity value as the abnormal window position, and reverse map to the corresponding device type and specific sensing channel according to the matrix row index and column index corresponding to the abnormal window position in a preset order to determine the fault device number and device type corresponding to the abnormal window.

[0012] Preferably, it also includes a health status trend prediction module, which includes: The time series acquisition submodule is used to acquire continuous sequences prior to the current monitoring period. The comprehensive health score and adaptive comprehensive health evaluation threshold of each power supply unit for each historical monitoring period are used to construct the time series sequence of the comprehensive health score and the time series sequence of the adaptive comprehensive health evaluation threshold for each power supply unit. Preset the history window length; The trend prediction submodule is used to input the time series sequence of the comprehensive health score of each power supply unit and the time series sequence of the adaptive comprehensive health evaluation threshold into the trained trend prediction model. The trend prediction model outputs the future... The predicted comprehensive health score sequence and the predicted adaptive comprehensive health evaluation threshold sequence for each power supply unit during each monitoring period. Preset prediction step size; The fault prediction submodule is used to predict future faults. For each monitoring period, the predicted comprehensive health score sequence and the predicted adaptive comprehensive health evaluation threshold sequence for each power supply unit are used. The difference between the predicted comprehensive health score and the predicted adaptive comprehensive health evaluation threshold is calculated for each future monitoring period and used as the predicted difference for that future monitoring period. Future monitoring periods with a predicted difference less than zero are selected as future fault risk periods. When the number of future fault risk periods is greater than or equal to... If so, a manual investigation and warning will be triggered.

[0013] Preferably, the visual interactive scheduling module includes: The fault equipment highlighting and positioning submodule is used to query the standard 3D model component corresponding to the fault equipment in the power supply BIM model based on the fault equipment number and equipment type output by the fault location and tracing module, highlight the standard 3D model component in the 3D visualization scene with a preset highlight color, and automatically focus the 3D scene view to the 3D spatial coordinate position of the fault equipment. The spare parts location retrieval submodule is used to send an inventory query request to the enterprise material management system through a standard API interface based on the spare parts material code of the faulty equipment. It receives the inventory quantity and three-dimensional spatial coordinates of the spare parts storage location of each work area's spare parts warehouse, filters out the spare parts storage locations with an inventory quantity greater than zero, calculates the spatial distance between each spare parts storage location and the faulty equipment, determines the spare parts storage location with the smallest spatial distance as the target spare parts location, and outputs the three-dimensional spatial coordinates of the spare parts storage location and the warehouse shelf information. The optimal maintenance personnel matching submodule is used to match the corresponding maintenance professional category according to the equipment type of the faulty equipment. Based on the maintenance professional category, the maintenance personnel's shift list and the maintenance personnel's skill list, it filters out the maintenance personnel who are currently on duty and have the corresponding professional skills, and sorts them in a priority order from high to low according to the preset skill level to determine the optimal maintenance personnel. The maintenance work order generation and push submodule is used to push maintenance work orders, which include the name of the faulty equipment, the three-dimensional spatial coordinates of the faulty equipment, the defect level, the optimal maintenance strategy, the coordinates of the spare parts storage location, the spare parts warehouse shelf information, and the contact information of the best maintenance personnel, to the handheld maintenance terminal of the railway power supply equipment of the best maintenance personnel via the railway private network using the HTTPS protocol.

[0014] The beneficial effects of this invention compared to the prior art are as follows: This invention solves the problem of a lengthy information transmission chain in manual inspections by setting up a power supply BIM model construction module and a visualization and interactive scheduling module to directly push the three-dimensional coordinates of faulty equipment to the terminal of maintenance personnel. By setting up a power supply unit data acquisition and fusion module and a fault location and tracing module, a fused time-series status feature matrix that includes substation equipment, feeder cables, and contact network equipment is constructed, realizing the correlation analysis of upstream and downstream equipment within the same power supply unit and solving the problem of fault location deviation caused by the lack of inter-equipment transmission correlation analysis in single-equipment monitoring methods. By setting up a power supply health status assessment module, a predictive pantograph status index is introduced to adaptively adjust the comprehensive health evaluation threshold, so that the evaluation standard dynamically changes with the actual state of the pantograph, solving the problem of the disconnect between fixed evaluation standards and operating conditions.

[0015] Other features and advantages of the invention will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in this application.

[0016] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0017] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings: Figure 1 This is a schematic diagram of a BIM-based railway power supply equipment operation and maintenance platform according to an embodiment of the present invention. Detailed Implementation

[0018] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention. Example 1:

[0019] refer to Figure 1 This invention provides a BIM-based railway power supply equipment operation and maintenance platform, comprising: The power supply BIM model building module is used to build power supply BIM models of each power supply unit along the railway line based on the standardized railway power supply BIM component family library, 3D spatial drawings of railway power supply equipment and electrical connection diagrams of railway power supply equipment. The power supply unit data acquisition and fusion module is used to collect time-series data of various types of substation equipment, various types of feeder cables and various types of contact network equipment of each power supply unit in each monitoring period, and fuse them to obtain the fused time-series status feature matrix of each power supply unit in each monitoring period. The power supply health status assessment module is used to obtain the comprehensive health score of each power supply unit in each monitoring period based on the fused temporal state feature matrix of each power supply unit and the trained comprehensive health score model. By adaptively adjusting the comprehensive health evaluation threshold of the current monitoring period through the predicted pantograph state index when the train runs to the current power supply unit to be evaluated in each monitoring period, the module obtains the adaptive comprehensive health evaluation threshold for each monitoring period. Based on the comprehensive health score of each power supply unit in each monitoring period and the adaptive comprehensive health evaluation threshold, the module outputs the power supply health status assessment result of each power supply unit in each monitoring period. The fault location and tracing module is used to trace and locate faults in equipment based on the fusion time-series state feature matrix of each power supply unit and the corresponding power supply BIM model for each monitoring period, to determine the defect level of the faulty equipment and to provide the optimal maintenance strategy. The visual interactive scheduling module is used to visually identify and locate faulty equipment in the power supply BIM model based on the fault location and tracing results, retrieve the inventory location of the corresponding spare parts for the faulty equipment, match the best maintenance personnel, generate maintenance work orders and push them to the maintenance personnel's terminals.

[0020] In this embodiment, the standardized railway power supply BIM component family library refers to a pre-built collection of components that covers standard three-dimensional models of various equipment in the railway power supply system. Each standard three-dimensional model component in the component collection includes the corresponding equipment's external dimensions, equipment type, equipment model, equipment number, and spare parts material code.

[0021] In this embodiment, the three-dimensional spatial drawing of railway power supply equipment refers to the set of X-axis, Y-axis, and Z-axis coordinate data of each power supply unit, each feeder cable and its accessories, and each contact network equipment in three-dimensional space, obtained through on-site surveying.

[0022] In this embodiment, the railway power supply equipment electrical connection diagram refers to a data set that records the electrical connection relationships between electrical components and equipment in each power supply unit along the railway through feeder cables, busbars, and conductors. Each electrical connection relationship record includes the starting device identifier, the ending device identifier, and the connection type.

[0023] In this embodiment, a power supply unit refers to a power supply system unit consisting of a traction substation, all feeder cables of the substation, and all contact network sections powered by these feeder cables. A railway line may contain multiple power supply units, which are electrically isolated from each other through phase separation sections.

[0024] In this embodiment, the power supply BIM model includes the three-dimensional spatial coordinates of each device, the directed connection relationships between devices, device number, device type, spare parts material code, and the maintenance area information where the device is located. The direction of the directed connection relationship is from the traction transformer in the traction substation to the feeder cable, and from the feeder cable to the contact wire. This direction is consistent with the physical transmission direction of current flowing from the substation to the contact wire.

[0025] In this embodiment, the time-series data of the power equipment refers to the data sequence formed by arranging the operating status data of the key equipment in the substation that is prone to failure in a continuous monitoring time according to time order. Specifically, it includes the time-series data of traction transformer oil temperature, the time-series data of dissolved acetylene concentration in traction transformer oil, the time-series data of dissolved hydrogen concentration in traction transformer oil, the time-series data of grounding current of traction transformer core, the time-series data of partial discharge in GIS, the time-series data of vibration spectrum of GIS cabinet, the time-series data of SF6 gas concentration, and the time-series data of leakage current of surge arrester.

[0026] In this embodiment, the feeder cable time-series data refers to the data sequence formed by arranging the operating status data of the feeder cable and its accessories in chronological order during continuous monitoring. Specifically, it includes the time-series data of the distributed optical fiber temperature of the feeder cable sheath, the time-series data of the feeder cable grounding leakage current, and the time-series data of the feeder cable insulation loss coefficient.

[0027] In this embodiment, the contact network equipment time sequence data refers to the data sequence formed by arranging the operating status data of key equipment on the contact network that are prone to failure in a continuous monitoring time according to time order. Specifically, it includes contact network dropper tension time sequence data, contact network cantilever vibration amplitude time sequence data, contact network pantograph contact pressure time sequence data, contact network insulator leakage current time sequence data, contact network conductor wear thickness time sequence data, and contact network catenary tension time sequence data.

[0028] In this embodiment, the fused time-series state feature matrix is ​​a two-dimensional matrix formed by concatenating the time-series vectors of the substation equipment, feeder cable, and contact network equipment of the same power supply unit within the same monitoring period along the feature dimension and stacking them in chronological order. Each row of this matrix corresponds to a monitoring time, and each column corresponds to a sensing channel, with rows arranged in chronological order. Among them, the time sequence vector of power equipment refers to a one-dimensional vector formed by arranging the sensor data of all types of power equipment at the same monitoring time within the same monitoring period according to the preset sensor channel order. The feeder cable time sequence vector refers to a one-dimensional vector formed by arranging the sensor data of all types of feeder cables at the same monitoring time within the same monitoring period according to the preset sensor channel order. The catenary equipment time sequence vector refers to a one-dimensional vector formed by arranging the sensor data of all types of catenary equipment at the same monitoring time within the same monitoring period according to the preset sensor channel order.

[0029] In this embodiment, the comprehensive health score model is a deep learning model trained on a large number of historical fusion time-series state feature matrices and corresponding expert-annotated comprehensive health scores. This model maps the input fusion time-series state feature matrix to a comprehensive health score within the range of 0 to 1. Specifically: The model employs a Long Short-Term Memory (LSTM) network structure. The input layer receives a fused temporal state feature matrix, the hidden layer consists of two LSTM layers, each containing 128 neurons, and the output layer is a fully connected layer, outputting a comprehensive health score. The training method involves collecting a large number of fused temporal state feature matrix samples from each power supply unit during historical monitoring periods. Experts assign a comprehensive health score of 0 to 1 to each sample based on actual equipment maintenance records. The network parameters are trained using the mean squared error loss function and stochastic gradient descent algorithm.

[0030] In this embodiment, the pantograph condition index is a continuous value between 0 and 1, derived from three dimensions: pantograph carbon sliding plate wear, pantograph-catenary contact pressure, and pantograph lifting time. A value closer to 1 indicates a better pantograph condition, while a value closer to 0 indicates a worse condition. This index is a comprehensive quantitative indicator used to assess the overall health of the pantograph. Carbon sliding plate wear reflects the degree of mechanical wear between the pantograph and the contact wire; pantograph-catenary contact pressure reflects the quality of electrical contact between the pantograph and the contact wire; and pantograph lifting time reflects the working state of the pantograph's mechanical transmission mechanism.

[0031] In this embodiment, the predicted pantograph state index when the train reaches the current power supply unit to be evaluated is a parameter used to quantitatively evaluate the performance state of the pantograph at a future power supply unit. This value comprehensively reflects the expected state of the pantograph's carbon skid plate wear, pantograph-catenary contact pressure, and pantograph lifting time at a future moment.

[0032] In this embodiment, the comprehensive health score of each power supply unit in each monitoring period refers to the comprehensive health score value of the power supply unit at the last monitoring moment of that period. The value ranges from 0 to 1, with a closer value to 1 indicating a better health status and a closer value to 0 indicating a worse health status. The adaptive comprehensive health evaluation threshold is a threshold obtained by adaptively adjusting the basic comprehensive health evaluation threshold based on the predicted pantograph status index. It also ranges from 0 to 1 and is used to compare with the comprehensive health score to determine whether the health status of the power supply unit meets the standard.

[0033] In this embodiment, the power supply health status assessment results include the comprehensive health score and comprehensive health status label of the power supply unit. The comprehensive health status label includes four health status levels: A, B, C, and D. Level A indicates the best health status, and level D indicates the worst health status.

[0034] The beneficial effects of the above technologies are as follows: The power supply BIM model construction module of this invention constructs power supply BIM models for each power supply unit along the railway line based on a standardized railway power supply BIM component family library, 3D spatial drawings of railway power supply equipment, and electrical connection diagrams of railway power supply equipment. This power supply BIM model includes the 3D spatial coordinates, equipment number, equipment type, and spare parts material code of each piece of equipment, providing a spatial data foundation for the precise location of faulty equipment. The visual interactive scheduling module visually identifies and locates faulty equipment in the power supply BIM model based on the fault location and tracing results, retrieves the inventory location of the corresponding spare parts for the faulty equipment, matches the best maintenance personnel, generates maintenance work orders, and pushes them to the maintenance personnel's terminals. Through the collaborative operation of the above two modules, the 3D spatial coordinates of the faulty equipment are directly pushed to the handheld terminals of maintenance personnel, eliminating multiple intermediate steps in the manual inspection method, such as the hierarchical reporting of fault information, manual querying of duty rosters by the dispatching department, and manual notification of work area personnel. This minimizes the time delay between fault discovery and maintenance response, significantly improving emergency response efficiency. The power supply unit data acquisition and fusion module collects time-series data of various types of substation equipment, feeder cables, and overhead contact line equipment from each power supply unit during each monitoring period, and merges them to obtain a fused time-series status feature matrix for each power supply unit during each monitoring period. This fused time-series status feature matrix aligns the sensor data of the three physical levels of substation equipment, feeder cables, and overhead contact line equipment within the same power supply unit on a unified time axis and splices them in the feature dimension, so that the operating data of all key equipment within the same power supply unit can be comprehensively analyzed under a unified data framework. The fault location and tracing module performs fault location and tracing of faulty equipment based on the fused time-series state feature matrix of each power supply unit and the corresponding power supply BIM model for each monitoring period. When an abnormality occurs at a certain measuring point (such as downstream contact network equipment), the fused time-series state feature matrix of the measuring point contains the sensing data of upstream substation equipment and feeder cable in the same power supply unit at the same time. The fault location and tracing module can determine the faulty equipment by comprehensively analyzing the data of all equipment in the same power supply unit, avoiding the conductive interference caused by the upstream equipment fault being transmitted to the measuring point along the conduction path, and tracing the true root cause of the fault. This effectively solves the problem of fault location deviation and invalid inspection caused by the lack of inter-equipment correlation analysis in the existing technology. The power supply health status assessment module obtains the comprehensive health score of each power supply unit in each monitoring period based on the fused temporal state feature matrix of each power supply unit in each monitoring period and the trained comprehensive health score model. It also adaptively adjusts the comprehensive health evaluation threshold of the current monitoring period by using the predicted pantograph state index when the train runs to the current power supply unit to be evaluated in each monitoring period, thus obtaining the adaptive comprehensive health evaluation threshold for each monitoring period. Based on the comprehensive health score of each power supply unit in each monitoring period and the adaptive comprehensive health evaluation threshold, the module outputs the power supply health status assessment result of each power supply unit in each monitoring period. Through the above mechanism, when the predicted pantograph condition index is low, it indicates severe wear of the pantograph's carbon skid plate, deviation of the pantograph-catenary contact pressure from the standard value, or abnormal pantograph lifting time. This indicates a higher demand for power supply quality from the pantograph. In this case, the adaptive comprehensive health evaluation threshold automatically increases, and the evaluation standard becomes stricter, enabling timely identification of the mismatch risk between the actual power supply capacity of the power supply unit and the actual power demand of the pantograph. Conversely, when the predicted pantograph condition index is high, the adaptive comprehensive health evaluation threshold automatically decreases, and the evaluation standard becomes more lenient, avoiding over-evaluating power supply units that actually meet power supply requirements as unhealthy and triggering unnecessary maintenance procedures. By introducing the predicted pantograph condition index to adaptively adjust the evaluation threshold, the health evaluation standard of the power supply unit can dynamically change with the predicted state of the pantograph, matching the evaluation results with actual operating conditions. This solves the problem of evaluation result deviation caused by the disconnect between fixed evaluation standards and operating conditions in existing technologies, ensuring driving safety and avoiding unnecessary waste of maintenance resources. Example 2:

[0035] Based on Example 1, the power supply BIM model building module includes: The component family library acquisition submodule is used to acquire standard 3D model components of each power supply unit, each substation equipment, each feeder cable and its accessories, and each contact network equipment in each power supply unit along the railway line, based on the standardized railway power supply BIM component family library. Each standard 3D model component contains the corresponding equipment type and the corresponding equipment model. The 3D spatial coordinate acquisition submodule is used to acquire the actual 3D spatial coordinates of each power supply unit, each feeder cable and its accessories, and each contact network equipment within the railway line based on the 3D spatial drawings of the railway power supply equipment, as well as the spatial location coordinates and corresponding quantities of spare parts for each power supply unit, each feeder cable and its accessories, and each contact network equipment. The electrical connection relationship acquisition submodule is used to acquire the electrical connection relationships between electrical components and equipment in each power supply unit along the railway line based on the electrical connection diagram of the railway power supply equipment. The BIM model assembly generation submodule is used to take each standard 3D model component as the basic component unit, place each basic component unit in the corresponding spatial position according to the 3D spatial coordinates of each electrical component and equipment, establish directed connection edges between each basic component unit according to the electrical connection relationship between each electrical component and equipment, and generate the power supply BIM model of each power supply unit along the railway line.

[0036] In this embodiment, each electrical component includes each transformer equipment in each power supply unit, each feeder cable and its accessories, and each contact network equipment.

[0037] In this embodiment, the spatial coordinates of spare parts refer to the three-dimensional spatial coordinates of the storage location of each spare part in the spare parts warehouse of each work area, including warehouse number, shelf number, layer number, compartment number and corresponding X-axis, Y-axis and Z-axis coordinate values, which are used to quickly locate spare parts during subsequent maintenance.

[0038] In this embodiment, the specific rules for generating directed connection edges in the BIM model assembly generation submodule are as follows: taking the traction transformer in the traction substation as the starting node, determining each intermediate node sequentially along the direction of the feeder cable, and finally taking the end of the contact wire anchor section as the termination node. Each directed edge is represented by a triple (starting device identifier, termination device identifier, connection type).

[0039] The beneficial effects of the above technologies are as follows: This invention ensures that the three-dimensional spatial coordinates of each device in the BIM model are strictly consistent with the actual installation position, and that the direction of the directed connection edge is strictly consistent with the physical transmission direction of the current flowing from the traction substation to the contact network. It provides accurate spatial coordinate data for highlighting and locating faulty equipment in the three-dimensional scene, provides spare parts storage location coordinate data for spare parts location retrieval, provides directed connection relationship data for fault tracing, and provides a complete and accurate BIM model data foundation for the subsequent implementation of the fault location tracing module and the visualization interactive scheduling module. Example 3:

[0040] Based on Example 1, the power supply unit data acquisition and fusion module includes: The substation data acquisition submodule is used to collect data on the traction transformer oil temperature, dissolved acetylene concentration in the traction transformer oil, dissolved hydrogen concentration in the traction transformer oil, grounding current of the traction transformer core, partial discharge quantity inside the GIS, vibration spectrum of the GIS cabinet, SF6 gas concentration, and leakage current of the surge arrester in each power supply unit at each monitoring time during each monitoring period. It also arranges the same type of sensor data at all monitoring times within the same monitoring period in chronological order to form the time sequence data of each type of substation equipment in each monitoring period, and arranges the sensor data of all types of substation equipment at the same monitoring time within the same monitoring period in the order of the preset sensor channels to form the time sequence vector of the substation equipment at the corresponding monitoring time of the corresponding monitoring period. The feeder data acquisition submodule is used to collect sensor data on the distributed optical fiber temperature of the feeder cable sheath, the feeder cable grounding leakage current, and the feeder cable insulation loss coefficient of the feeder cable in each power supply unit at each monitoring time during each monitoring period. It also arranges the same type of sensor data at all monitoring times within the same monitoring period in chronological order to form the time sequence data of each type of feeder cable for each monitoring period. Finally, it arranges the sensor data of all types of feeder cables at the same monitoring time within the same monitoring period in the order of the preset sensor channels to form the feeder cable time sequence vector for the corresponding monitoring time of the corresponding monitoring period. The overhead contact line data acquisition submodule is used to collect data on the overhead contact line equipment in each power supply unit at each monitoring time during each monitoring period, including the tension of the overhead contact line droppers, the vibration amplitude of the overhead contact line brackets, the contact pressure of the overhead contact line pantograph, the leakage current of the overhead contact line insulators, the wear thickness of the overhead contact line conductors, and the tension of the overhead contact line catenary. It also arranges the same type of sensor data at all monitoring times within the same monitoring period in chronological order to form the chronological data of each type of overhead contact line equipment in each monitoring period. Finally, it arranges the sensor data of all types of overhead contact line equipment at the same monitoring time within the same monitoring period in the order of the preset sensor channels to form the chronological vector of the overhead contact line equipment at the corresponding monitoring time of the corresponding monitoring period. The time-series feature fusion submodule is used to concatenate the time-series vectors of substation equipment, feeder cable, and contact network equipment of the same power supply unit at the same monitoring time in the same monitoring period and at the same monitoring moment in the feature dimension to generate a fused time-series state feature row vector for each monitoring moment. The fused time-series state feature row vectors of all monitoring moments in the monitoring period are stacked in time sequence to obtain the fused time-series state feature matrix of each power supply unit in each monitoring period.

[0041] In this embodiment, the data acquisition methods for traction transformer oil temperature, dissolved acetylene concentration in traction transformer oil, dissolved hydrogen concentration in traction transformer oil, traction transformer core grounding current, partial discharge quantity inside the GIS, vibration spectrum of the GIS cabinet, SF6 gas concentration, and surge arrester leakage current sensing data are as follows: The oil temperature is collected using a fiber optic temperature sensor; the dissolved acetylene and hydrogen concentrations in the oil are collected using an online oil chromatography monitoring device; and the core grounding current is collected using a through-type current transformer. The partial discharge quantity inside the GIS is collected using an ultra-high frequency partial discharge sensor; the cabinet vibration spectrum is collected using a piezoelectric accelerometer; and the SF6 gas concentration is collected using an infrared absorption SF6 sensor. The surge arrester leakage current is collected using a leakage current monitor. The installation locations and data collection methods of the above sensors utilize mature technologies in the railway power supply field. The sampling frequency of each sensor is no less than 0.5Hz, and the data is uploaded to the platform data center in real time via a dedicated railway communication network.

[0042] In this embodiment, the sensor data acquisition methods for the distributed optical fiber temperature of the feeder cable sheath, the feeder cable grounding leakage current, and the feeder cable insulation loss coefficient are as follows: Among them, the temperature of the distributed optical fiber on the outer sheath of the feeder cable is collected by a distributed optical fiber temperature measurement system laid along the outer wall of the cable, the grounding leakage current is collected by a through-type current transformer installed on the grounding wire of the cable, and the insulation loss coefficient is collected by an online monitoring device for dielectric loss.

[0043] In this embodiment, the data acquisition methods for the contact wire suspension tension, contact wire bracket vibration amplitude, contact wire pantograph-catenary contact pressure, contact wire insulator leakage current, contact wire wear thickness, and contact wire catenary tension sensing data are as follows: Among them, the tension of the dropper is collected by a tension sensor, the vibration amplitude of the cantilever arm is collected by a piezoelectric accelerometer, the contact pressure of the pantograph and catenary is collected by a pressure sensor, the leakage current of the insulator is collected by a leakage current monitor, the wear thickness of the conductor is collected by a linear array wear camera, and the tension of the catenary cable is collected by a tension sensor.

[0044] In this embodiment, a monitoring period is defined. Contains At each monitoring time, then: No. The monitoring time of the first monitoring moment The time sequence vector of the substation equipment corresponding to each power supply unit is: ,in, The total number of sensing channels for power equipment (including sensing channels corresponding to traction transformer oil temperature, dissolved acetylene concentration in traction transformer oil, dissolved hydrogen concentration in traction transformer oil, grounding current of traction transformer core, partial discharge quantity inside GIS, vibration spectrum of GIS cabinet, SF6 gas concentration, and leakage current sensing data of surge arrester). For the first The power supply unit is the first The monitoring time of the first monitoring moment Normalized sensing values ​​of each transformer equipment sensing channel; No. The monitoring time of the first monitoring moment The timing vector of the feeder cable corresponding to each power supply unit is ,in, This refers to the total number of sensing channels for the feeder cable (including the sensing channels corresponding to the sensing data of distributed optical fiber temperature of the feeder cable sheath, grounding leakage current of the feeder cable, and insulation loss coefficient of the feeder cable). For the first The power supply unit is the first The monitoring time of the first monitoring moment Normalized sensing values ​​for each feeder cable sensing channel; No. The monitoring time of the first monitoring moment The timing vector of the overhead contact line equipment corresponding to each power supply unit is: ,in, The total number of sensing channels for the overhead contact system equipment (including the sensing channels corresponding to the sensing data of overhead contact line dropper tension, overhead contact line cantilever vibration amplitude, overhead contact line pantograph contact pressure, overhead contact line insulator leakage current, overhead contact line conductor wear thickness, and overhead contact line catenary tension). For the first The power supply unit is the first The monitoring time of the first monitoring moment Normalized sensing values ​​of each overhead contact line device sensing channel.

[0045] In this embodiment, the monitoring period No. The fusion timing state feature matrix corresponding to each power supply unit Represented as: ; in, for OK The matrix consists of columns, each row corresponding to a monitoring time, and each column corresponding to a sensing channel, with the rows arranged in chronological order.

[0046] The beneficial effects of the above technology are as follows: The sensor data from the three physical levels—transformer equipment, feeder cables, and overhead contact line equipment—are aligned on a unified time axis and organically integrated in the feature dimension. This results in each row of the final fused time-series state feature matrix containing data from all key sensors of the transformer equipment, feeder cables, and overhead contact line equipment within the same power supply unit at the same monitoring time. When an anomaly occurs at a measurement point of a certain device (such as downstream overhead contact line equipment), the fused time-series state feature matrix simultaneously contains sensor data from upstream transformer equipment and feeder cables within the same power supply unit at the same time. This provides complete data support for the fault location and tracing module to distinguish whether the anomaly is caused by the device itself or by upstream transmission, effectively solving the technical deficiency in existing technologies where independent data collection, storage, and analysis by each device prevent inter-device correlation analysis. Example 4:

[0047] Based on Example 1, the power supply health status assessment module includes: The health score acquisition submodule is used to input the fused time-series status feature matrix of each power supply unit in each monitoring period into the trained comprehensive health score model, and output the comprehensive health score of each power supply unit through the trained comprehensive health score model. The pantograph status evaluation submodule is used to collect various types of operating status data of the train pantograph during each monitoring period, and calculate the current pantograph status index of the train pantograph during each monitoring period based on the remaining thickness of the pantograph carbon sliding plate, the pantograph-catenary contact pressure and the pantograph lifting action time in the various types of operating status data. The adaptive evaluation threshold determination submodule is used to calculate the comprehensive health evaluation threshold for each monitoring period based on the pantograph status index and the basic comprehensive health evaluation threshold of the train pantograph for each monitoring period. The status assessment result judgment submodule is used to calculate the difference between the comprehensive health score of each power supply unit and the adaptive comprehensive health evaluation threshold for each monitoring period. Based on the preset difference range into which the difference falls, the power supply health status assessment level corresponding to each power supply unit for each monitoring period is output.

[0048] In this embodiment, the various types of operating status data include the remaining thickness of the pantograph carbon skid plate, the pantograph-catenary contact pressure data, and the pantograph lifting action time data.

[0049] In this embodiment, the remaining thickness of the pantograph carbon contact plate is collected in real time by laser displacement sensors installed at both ends of the pantograph carbon contact plate. The pantograph-catenary contact pressure is collected in real time by a pressure sensor installed between the pantograph contact plate bracket and the pantograph head. The pantograph raising action time is collected in real time by an angle sensor installed at the hinge of the pantograph base. The angle sensor detects the angle between the pantograph base and the horizontal plane. The starting moment of the raising action is defined as the moment when the angle begins to increase from the rest position, and the completion moment of the raising action is defined as the moment when the angle increases to 45 degrees (the standard working position where the pantograph contact plate contacts the catenary wire). The time difference between the starting moment and the completion moment is the raising action time.

[0050] In this embodiment, the current pantograph state index The calculation method is as follows: First, calculate the wear score of the carbon skateboard. ,in, The remaining thickness of the pantograph carbon slide plate. The new pantograph carbon skid plate has a standard thickness. Calculate the pantograph-catenary contact force score ,in, For the contact pressure of the bow and catenary, This is the recommended value for standard pantograph-catenary contact force; Calculate the bow lift time score ,in, For the time of the bow raising action, Standard bow raising time; Then, the current pantograph state index is calculated by weighted summation. ,in ,default , , ; The value can be adjusted manually based on actual engineering needs.

[0051] In this embodiment, the difference between the comprehensive health score of each power supply unit and the adaptive comprehensive health evaluation threshold during each monitoring period is... ,in, Assess the overall health of the power supply unit. The adaptive comprehensive health evaluation threshold for the power supply unit; The preset difference range is: Corresponding to Grade A; Corresponding to Grade B; Corresponding to Level C; Corresponding to Level D; In this rating system, Grade A indicates the best health status of the power supply unit, Grade B indicates good health status, Grade C indicates poor health status, and Grade D indicates the worst health status. When the evaluation grade is Grade C or D, the power supply unit is marked as a unit to be inspected, triggering the fault location and tracing module to perform fault diagnosis.

[0052] The beneficial effects of the above technologies are as follows: The health score acquisition submodule of this invention inputs the fused temporal state feature matrix of each power supply unit in each monitoring period into the trained comprehensive health score model, and outputs the comprehensive health score of each power supply unit through the trained comprehensive health score model, realizing end-to-end automated health assessment. The pantograph status evaluation submodule collects various types of operating status data of the train pantograph in each monitoring period, and calculates the current pantograph status index of the train pantograph in each monitoring period based on the remaining thickness of the pantograph carbon sliding plate, the pantograph-catenary contact pressure, and the pantograph lifting action time in each type of operating status data. This status index comprehensively reflects the mechanical wear degree of the pantograph carbon sliding plate, the quality of the pantograph-catenary electrical contact, and the working status of the pantograph lifting transmission mechanism. The adaptive evaluation threshold determination submodule calculates the comprehensive health evaluation threshold for each monitoring period based on the current pantograph status index of the train pantograph in each monitoring period and the basic comprehensive health evaluation threshold. When the pantograph status index is low, the adaptive comprehensive health evaluation threshold automatically increases and the evaluation standard becomes stricter; when the pantograph status index is high, the adaptive comprehensive health evaluation threshold automatically decreases and the evaluation standard becomes more lenient. The status assessment result judgment submodule calculates the difference between the comprehensive health score of each power supply unit and the adaptive comprehensive health evaluation threshold for each monitoring period, and outputs the power supply health status assessment level corresponding to each power supply unit for each monitoring period based on the preset difference range into which the difference falls. Through the sequential coordination of the aforementioned sub-modules, the current pantograph status index output by the pantograph status evaluation sub-module is used by the adaptive evaluation threshold determination sub-module to dynamically adjust the comprehensive health evaluation threshold. The status assessment result judgment sub-module then compares the adjusted threshold with the comprehensive health score to output the evaluation level. This allows the same power supply unit to use different evaluation standards when supplying power to pantographs in different states. When the pantograph's carbon sliding plate is severely worn or the pantograph-catenary contact pressure is abnormal, the requirements for power supply quality are higher, and the system automatically raises the evaluation threshold to ensure that the evaluated power supply unit can achieve stable and high-quality power supply. This avoids the problem that the power supply unit itself is fault-free, but the power supply quality decreases when it is paired with a pantograph in poor condition, thus affecting the normal operation of the train. When the pantograph is in good condition, the evaluation threshold remains at the basic level to avoid excessive maintenance. This solves the technical problem in the prior art where the evaluation results are out of sync with the actual operating conditions because the fixed evaluation standards cannot be adaptively adjusted according to the actual state of the pantograph. Example 5:

[0053] Based on Example 4, the adaptive evaluation threshold determination submodule includes: The train status data acquisition unit is used to obtain the average speed of the train during its operation, which is used as the average speed of the train. It also obtains the distance from the current position of the train's pantograph to the corresponding power supply unit at the end of each monitoring period, which is used as the distance the train needs to travel in each monitoring period. The state index prediction execution unit is used to input the train's average driving speed, the train's distance to be traveled in each monitoring period, and the current pantograph state index of the train's pantograph in each monitoring period into the trained pantograph state index prediction model to obtain the predicted pantograph state index when the train runs to the current power supply unit to be evaluated. The adaptive adjustment coefficient calculation unit is used to calculate the adaptive adjustment coefficient for the corresponding monitoring period based on the predicted pantograph state index when the train runs to the current power supply unit to be evaluated. The adaptive evaluation threshold calculation unit is used to calculate the comprehensive health evaluation threshold for each monitoring period based on the basic comprehensive health evaluation threshold and the adaptive adjustment coefficient of the corresponding monitoring period.

[0054] In this embodiment, the average train speed is obtained by continuously sampling and averaging data from the train speed sensor as the train travels along the line, with the unit being kilometers per hour. The distance to be traveled is calculated by the difference between the mileage marker at the train's current location and the mileage marker at the end of the current power supply unit to be evaluated, with the unit being kilometers.

[0055] In this embodiment, the trained pantograph state index prediction model is a three-input single-output regression model built on a multilayer perceptron neural network. The three inputs are the average train speed. Distance to be traveled and current pantograph status index The output is the predicted pantograph state index. The training method for this model is as follows: A large amount of historical data on the average train speed, distance to be traveled, current pantograph state index, and the actual pantograph state index when the train arrives at the target power supply unit are collected. The first three are used as input features, and the latter as the output label. A multilayer perceptron neural network is used for training, with two hidden layers, each containing 64 neurons. The ReLU activation function is used, and the Sigmoid function is used in the output layer to limit the output to between 0 and 1. The training objective is to minimize the mean squared error between the model output and the label.

[0056] In this embodiment, when the train's current position has passed the current power supply unit to be evaluated, the remaining travel distance is zero, and at this time the pantograph state index is predicted. Directly equal to the current pantograph state index The prediction model will no longer be called for prediction.

[0057] In this embodiment, the adaptive adjustment coefficient for the corresponding monitoring period is calculated based on the predicted pantograph state index when the train reaches the current power supply unit to be evaluated. The calculation formula is: ; in, To predict the pantograph status index (value range 0 to 1). The preset maximum adjustment range coefficient (default 0.3, value range 0.2 to 0.5). The preset sensitivity coefficient is set (default 2.0, range 1.5 to 3.0); when predicting the pantograph state index... When it is high, the adaptive adjustment coefficient Approaching 1, the adaptive comprehensive health assessment threshold remains essentially unchanged; when predicting the pantograph state index... When the value is low, the adaptive adjustment coefficient A value significantly greater than 1 indicates an increase in the adaptive comprehensive health assessment threshold and a more stringent assessment standard. When the same power supply unit supplies power to multiple pantographs of multiple trains during the same monitoring period, state index prediction and adaptive adjustment coefficient calculation are performed for each pantograph, and the maximum value among all adaptive adjustment coefficients is taken as the adaptive adjustment coefficient of the power supply unit.

[0058] In this embodiment, the comprehensive health evaluation threshold for each monitoring period is calculated based on the basic comprehensive health evaluation threshold and the adaptive adjustment coefficient for the corresponding monitoring period. The calculation formula is: ; in, The basic comprehensive health assessment threshold is a preset fixed value, which is the default value. The value ranges from 0 to 1; The function is used to limit the adaptive comprehensive health evaluation threshold to below 0.95, so as to avoid triggering invalid maintenance by judging all power supply units as unhealthy due to an excessively high threshold.

[0059] In this embodiment, the adaptive adjustment coefficient is a dimensionless coefficient used to adjust the magnitude of the basic comprehensive health assessment threshold.

[0060] In this embodiment, the basic comprehensive health assessment threshold is a preset fixed value, ranging from 0 to 1, by default. This value is determined based on industry standards, equipment maintenance procedures, and historical maintenance experience in the field of railway power supply equipment operation and maintenance. Specifically, through statistical analysis of comprehensive health score data of power supply units over a large number of historical monitoring periods, the optimal dividing point that can effectively distinguish between healthy and unhealthy states is selected as the basic threshold. System administrators can adjust this value according to the specific operational characteristics and safety requirements of the line.

[0061] The beneficial effects of the above technology are as follows: The average speed and distance to be traveled provided by the train status data acquisition unit of this invention, together with the current pantograph status index provided by the pantograph status evaluation submodule, are input into the status index prediction execution unit to obtain the predicted pantograph status index. This predicted value is used by the adaptive adjustment coefficient calculation unit to calculate the adaptive adjustment coefficient, and finally, the adaptive evaluation threshold calculation unit generates the adaptive comprehensive health evaluation threshold. Since it takes a certain amount of time for the train to travel from its current position to the power supply unit to be evaluated, the pantograph status may degrade due to continuous wear during this period. This embodiment uses the pantograph status index prediction model to predict the status index of the pantograph when it arrives at the power supply unit to be evaluated in advance and adjusts the evaluation threshold accordingly. This makes the adjustment of the evaluation threshold proactive rather than reactive, avoiding evaluation delays caused by sudden deterioration of the pantograph status during train operation, and further improving the ability to ensure train safety. Example 6:

[0062] Based on Example 1, the fault location and tracing module includes: The fault screening and judgment submodule is used to receive the power supply health status assessment results of each power supply unit in each monitoring period output by the power supply health status assessment module, and to screen out unhealthy power supply units whose comprehensive health score is lower than the adaptive comprehensive health evaluation threshold, mark them as power supply units to be repaired, and trigger the fault tracing process. The fault equipment tracing submodule is used to compare the standard fusion time-series status feature matrix of each power supply unit in a healthy state with the fusion time-series status feature matrix of the corresponding power supply unit in the current monitoring period, determine the position of the sliding window with the largest deviation from the standard fusion time-series status feature matrix in the current monitoring period, and back-map the sliding window position to the corresponding equipment type and sensing channel to determine the equipment number and equipment type of the faulty equipment. The fault equipment location submodule is used to query the three-dimensional spatial coordinates and power supply unit information of the fault equipment in the power supply BIM model based on the equipment number of the fault equipment, and determine the operation and maintenance work area corresponding to the fault equipment. The defect level determination submodule is used to calculate the defect level value of the faulty equipment based on the similarity value of the sliding window corresponding to the faulty equipment, and to determine the defect level of the faulty equipment based on the preset defect level range into which the defect level value falls. The maintenance strategy determination submodule is used to match the optimal maintenance strategy from a preset maintenance strategy knowledge base based on the equipment type and defect level of the faulty equipment.

[0063] In this embodiment, the standard fusion time-series state feature matrix The method for obtaining the data is as follows: Collect fused time-series status feature matrix samples from multiple monitoring periods during the historical operation of each power supply unit, where experts have marked the status as healthy. Then, calculate the average value element-by-element for all healthy status samples. Specifically, assume there are a total of... Each of the following health status samples is a health status sample. OK Column matrix Given the total number of sensing channels, the standard fusion time-series state feature matrix is... The Line 1 The column element value is: ; in, For the first The fused temporal state feature matrix of each health state sample. .

[0064] In this embodiment, the first to second elements of the fused temporal state feature matrix are... Column of strain gauge equipment The first sensor channel, the first to The corresponding feeder cable The first sensor channel, the first to List of corresponding overhead contact line equipment Each sensor channel. Once the column index range of the anomaly window is located, the corresponding device type and specific sensor channel can be determined based on the column interval to which that range belongs. Then, the specific monitoring time is determined by combining the row index of the fused time-series status feature matrix, and finally, the specific faulty device number is determined.

[0065] In this embodiment, the faulty equipment location submodule queries the three-dimensional spatial coordinates of the faulty equipment in the power supply BIM model based on the equipment number of the faulty equipment. The system includes information about the power supply unit to which the equipment belongs, and determines the corresponding maintenance work area based on that unit. Each power supply unit has its corresponding maintenance work area information pre-bound in the BIM model.

[0066] In this embodiment, the defect level determination submodule contains a defect level calculation function. Using sliding window similarity Input: Defect level numerical value The defect level calculation function It is a continuously decreasing monotonically decreasing function; the lower the similarity (the greater the difference), the higher the defect level value. Specifically: ; in, The similarity value of the sliding window corresponding to the faulty device, with a value range of [value range missing]. , It is a rounding function. The range of values ​​is The physical meaning of this formula is: it maps similarity from 0 to 1 to defect levels from 4 to 1, with lower similarity resulting in higher defect level values; The preset defect level range is: when This corresponds to a level one defect; when This corresponds to a level 2 defect; when This corresponds to a level three defect; when This corresponds to a level four defect.

[0067] In this embodiment, the maintenance strategy knowledge base is a structured database containing a three-element mapping relationship between equipment type, defect level, and maintenance strategy. Its data structure consists of equipment type, defect level, and maintenance strategy. The maintenance strategy includes three dimensions: maintenance method, maintenance time limit, and required resources.

[0068] Maintenance methods include four types: equipment replacement, overhaul, targeted maintenance, and enhanced monitoring. Maintenance timeframes include four levels: immediate (within 24 hours), emergency (within 72 hours), planned (within a month), and routine (within a quarter). Required resources include required spare parts material codes, required maintenance personnel professional categories, and estimated maintenance man-hours.

[0069] In this embodiment, the method for constructing the maintenance strategy knowledge base is as follows: for each combination of equipment type and each defect level, railway power supply experts pre-determine the corresponding maintenance methods, maintenance time limits and required resources based on equipment maintenance procedures and actual operation and maintenance experience, thus forming a complete maintenance strategy knowledge base.

[0070] In this embodiment, the specific method for the maintenance strategy determination submodule to match the optimal maintenance strategy is as follows: using the equipment type and defect level of the faulty equipment as joint query conditions, the corresponding maintenance strategy record is retrieved in the maintenance strategy knowledge base. If a unique matching record is found, the maintenance strategy is directly output. If multiple matching records are found, the one with the shortest maintenance time limit is output. If no matching record is found, the maintenance strategy corresponding to the next level of defect level of the same equipment type is output.

[0071] The beneficial effects of the above technology are as follows: The fault screening and judgment submodule of this invention receives the power supply health status assessment results of each power supply unit in each monitoring period output by the power supply health status assessment module, and filters out unhealthy power supply units whose comprehensive health score is lower than the adaptive comprehensive health evaluation threshold. These unhealthy power supply units are marked as power supply units to be repaired, and the fault tracing process is triggered, achieving accurate triggering of fault diagnosis and avoiding unnecessary fault analysis calculations on healthy power supply units. The fault equipment tracing submodule compares the standard fused time-series status feature matrix of each power supply unit in a healthy state with the fused time-series status feature matrix of the corresponding power supply unit in the current monitoring period, determines the position of the sliding window with the largest deviation from the standard fused time-series status feature matrix in the current monitoring period, and back-maps the sliding window position to the corresponding equipment type and sensing channel to determine the equipment number and equipment type of the faulty equipment. Because the fused temporal state feature matrix simultaneously contains data from three levels—transformer equipment, feeder cables, and overhead contact line equipment—when an anomaly occurs at a certain measuring point, the position of the abnormal data in the matrix can be accurately located through sliding window comparison. Then, based on the column interval corresponding to that position, the specific equipment type and sensing channel are determined, achieving precise fault location from matrix-level overall anomaly detection to equipment-level precise fault location. The fault equipment location submodule, based on the fault equipment number, queries the three-dimensional spatial coordinates and power supply unit information of the equipment in the power supply BIM model to determine the corresponding maintenance work area. The defect level judgment submodule calculates the defect level value of the fault equipment using a defect level calculation function based on the sliding window similarity value corresponding to the fault equipment. The defect level of the fault equipment is determined based on the preset defect level interval into which the defect level value falls. The maintenance strategy determination submodule matches the corresponding optimal maintenance strategy from a preset maintenance strategy knowledge base based on the equipment type and defect level of the fault equipment. This maintenance strategy knowledge base is a structured database containing a ternary mapping relationship between equipment type, defect level, and maintenance strategy. The maintenance strategy includes three dimensions: maintenance method, maintenance time limit, and required resources. Through the sequential coordination of the above sub-modules, the fusion time-series state feature matrix of the power supply unit to be inspected selected by the fault screening and judgment sub-module is used by the fault equipment tracing sub-module to locate the abnormal window position through sliding window comparison and reverse mapping to determine the faulty equipment. The fault equipment location sub-module queries the spatial coordinates and maintenance work area in the BIM model based on the faulty equipment number. The defect level judgment sub-module calculates the defect level based on the similarity of the sliding window. The maintenance strategy determination sub-module matches the optimal maintenance strategy based on the equipment type and defect level. This forms a complete automated diagnostic link from anomaly detection to root cause location, from defect level determination to maintenance strategy generation. This allows maintenance personnel to perform accurate maintenance based on the maintenance strategy automatically generated by the system without having in-depth fault diagnosis expertise, avoiding the problem of inconsistent maintenance strategies caused by differences in human experience. Example 7:

[0072] Based on Example 6, the fault equipment tracing submodule includes: The traversal unit determination unit is used to determine the size of the traversal unit of the sliding window based on the evaluation level of the power supply unit corresponding to the comprehensive health status label of the power supply unit in the power supply health status evaluation results. The traversal unit is a rectangular window, and its row height and column width are determined according to the evaluation level. The similarity traversal execution unit is used to synchronously traverse the fusion time-series state feature matrix and the corresponding standard fusion time-series state feature matrix of the power supply unit to be inspected during the current monitoring period, using the rectangular sliding window determined by the traversal unit as the traversal unit. Each traversal extracts the current sub-matrix of the sliding window coverage area in the fusion time-series state feature matrix and the standard sub-matrix at the same position in the standard fusion time-series state feature matrix, calculates the similarity between the current sub-matrix and the standard sub-matrix, and records the similarity value corresponding to each sliding window position. The source identification and judgment unit is used to compare the similarity values ​​of all sliding window positions, locate the sliding window position with the smallest similarity value as the abnormal window position, and reverse map to the corresponding device type and specific sensing channel according to the matrix row index and column index corresponding to the abnormal window position in a preset order to determine the fault device number and device type corresponding to the abnormal window.

[0073] In this embodiment, the size of the sliding window traversal unit is determined based on the evaluation level of the power supply unit to be inspected: When the assessment level is C , ; When the assessment grade is D , ; in, and The default is the preset base window size. , When the assessment level is D (poor health status of the power supply unit), it indicates that the overall condition of the power supply unit is poor, which may involve anomalies in multiple devices or multiple sensor channels. Therefore, a larger window is used to cover a wider area for anomaly localization. When the assessment level is C, the anomaly may be limited to a few devices or sensor channels, and a smaller window is used for fine localization. If the row height h of the sliding window determined according to the assessment level is greater than the total number of rows K of the fused time-series status feature matrix for the current monitoring period, then the window height h is adjusted to K. If the column width w is greater than the total number of columns D of the matrix, then the window width w is adjusted to D.

[0074] In this embodiment, the traversal step size of the sliding window in both the row and column directions is 1, that is, each time a row or column is slid, all possible positions are covered row by row and column by column; when the sliding window moves to the edge of the matrix and the remaining number of rows or columns is insufficient to form a complete window, the incomplete area is ignored and no similarity calculation is performed.

[0075] In this embodiment, the similarity between the current submatrix and the standard submatrix Calculate using the following formula: ; in, The number of rows involved in the sliding window. The number of columns involved in the sliding window. For the current submatrix, the th Line 1 The element values ​​of the column, For the standard submatrix Line 1 The element values ​​of the column. The closer the value is to 1, the more similar the two sub-matrices are; the smaller the value, the greater the difference. This involves calculating the average of the absolute differences between corresponding elements in the two sub-matrices (the fused temporal state feature matrix and the standard fused temporal state feature matrix). The larger this average value, the greater the difference.

[0076] The beneficial effects of the above technology are as follows: The traversal unit determination unit of this invention determines the size of the traversal unit of the sliding window based on the evaluation level of the power supply unit corresponding to the comprehensive health status label of the power supply unit in the power supply health status assessment results. The traversal unit is a rectangular window, and its row height and column width are determined according to the evaluation level. When the evaluation level is C... , When the evaluation level is D , The similarity traversal execution unit uses the rectangular sliding window determined by the traversal unit as the traversal unit. It synchronously traverses the fused time-series state feature matrix and the corresponding standard fused time-series state feature matrix of the power supply unit under maintenance during the current monitoring period. Each traversal extracts the current sub-matrix of the sliding window coverage area in the fused time-series state feature matrix and the standard sub-matrix at the same position in the standard fused time-series state feature matrix, calculates the similarity between the current sub-matrix and the standard sub-matrix, and records the similarity value corresponding to each sliding window position. The source identification and judgment unit compares the similarity values ​​of all sliding window positions, locates the sliding window position with the smallest similarity value as the abnormal window position, and maps it back to the corresponding device type and specific sensing channel according to a preset order based on the matrix row index and column index of the abnormal window position, determining the faulty device number and device type corresponding to the abnormal window. Through the sequential cooperation of the above units, the window size adaptively determined by the traversal unit based on the evaluation level directly determines the data range covered by each window when the similarity traversal execution unit performs sliding window traversal. The source identification and judgment unit then locates the abnormal window position based on the traversal results and back-maps to determine the faulty device. When the assessment level is D, it indicates that the overall condition of the power supply unit is poor, and the anomaly may involve multiple devices or multiple sensor channels. In this case, using a larger sliding window can ensure that the abnormal area is completely covered, avoiding the abnormal area being fragmented into multiple pieces due to an excessively small window, thus preventing effective identification. When the assessment level is C, it indicates that the anomaly is limited to a few devices or sensor channels. In this case, using a smaller sliding window for fine traversal can achieve high-precision anomaly location, avoiding the reduction in location accuracy due to the introduction of too much normal data by an excessively large window. Compared with the existing technology that uses a fixed-size window for traversal, the adaptive window mechanism of this invention can achieve optimal anomaly detection results under different health states. For minor faults, a small window ensures location accuracy, while for severe faults, a large window ensures complete coverage of the abnormal area. At the same time, by mapping the anomaly window position back to the device type and sensor channel, an interpretable mapping from matrix space to physical device space is achieved, replacing manual inspection and troubleshooting, improving the efficiency of source tracing and location, and providing a clear target for subsequent precise maintenance. Example 8:

[0077] Based on Example 1, a health status trend prediction module is also included, which includes: The time series acquisition submodule is used to acquire continuous sequences prior to the current monitoring period. The comprehensive health score and adaptive comprehensive health evaluation threshold of each power supply unit for each historical monitoring period are used to construct the time series sequence of the comprehensive health score and the time series sequence of the adaptive comprehensive health evaluation threshold for each power supply unit. Preset the history window length; The trend prediction submodule is used to input the time series sequence of the comprehensive health score of each power supply unit and the time series sequence of the adaptive comprehensive health evaluation threshold into the trained trend prediction model. The trend prediction model outputs the future... The predicted comprehensive health score sequence and the predicted adaptive comprehensive health evaluation threshold sequence for each power supply unit during each monitoring period. Preset prediction step size; The fault prediction submodule is used to predict future faults. For each monitoring period, the predicted comprehensive health score sequence and the predicted adaptive comprehensive health evaluation threshold sequence for each power supply unit are used. The difference between the predicted comprehensive health score and the predicted adaptive comprehensive health evaluation threshold is calculated for each future monitoring period and used as the predicted difference for that future monitoring period. Future monitoring periods with a predicted difference less than zero are selected as future fault risk periods. When the number of future fault risk periods is greater than or equal to... If so, a manual investigation and warning will be triggered.

[0078] In this embodiment, To preset the history window length, the default is... This indicates that historical data from the most recent 10 monitoring periods are used as input for trend prediction. The default prediction step size is set to the preset step size. This represents the predicted comprehensive health score and adaptive comprehensive health evaluation threshold for the next 5 monitoring periods.

[0079] In this embodiment, the comprehensive health score time series is constructed as follows: the comprehensive health score of each power supply unit in each historical monitoring period is used as the basis for the construction. Arranged in chronological order, forming a length of sequence The adaptive comprehensive health assessment threshold time series is constructed as follows: the adaptive comprehensive health assessment threshold of each power supply unit in each historical monitoring period is used as the basis for the construction. Arranged in chronological order, forming a length of sequence .

[0080] In this embodiment, the trend prediction model employs a long short-term memory network based on an encoder-decoder structure, with the encoder portion receiving a length of [missing information]. The comprehensive health score sequence and the adaptive comprehensive health evaluation threshold sequence (two sequences total) The numerical values ​​are encoded into a context vector; the decoder part decodes step by step based on this context vector, outputting the future value. Predicted comprehensive health score for each monitoring period And predictive adaptive comprehensive health assessment threshold ( The training method for this model is as follows: Collect comprehensive health scores and adaptive comprehensive health evaluation threshold data for each power supply unit during historical monitoring periods, and continuously... Data from a historical monitoring period was used as the input sample, followed by... The comprehensive health score and adaptive comprehensive health evaluation threshold for each monitoring period are used as output labels to construct a training dataset, and the mean squared error loss function is used to train the network.

[0081] In this embodiment, the predicted difference ,in The preset difference range is: Corresponding to Grade A, Corresponding to Grade B, Corresponding to Level C, Corresponding to Level D. Indicates to Round up, when hour If the predicted assessment level is C or D for 3 or more of the next 5 monitoring periods, the power supply unit is deemed to have a risk of future failure. A future monitoring period with a prediction difference of less than zero is considered a future monitoring period with a power supply health status assessment level of C or D.

[0082] The beneficial effects of the above technology are as follows: The time-series acquisition submodule of this invention organizes the discrete evaluation results of historical monitoring periods into sequence data with a time-series structure. The trend prediction submodule uses a long short-term memory network based on this sequence data to capture the long-term dependency relationship and nonlinear change pattern of the health score over time and outputs predicted values ​​for multiple future monitoring periods. The fault prediction submodule filters future fault risk periods based on the predicted values ​​and triggers an early warning when the risk period reaches a threshold. This embodiment can not only assess the current health status of the power supply unit, but also predict the trend of health status changes for multiple future monitoring periods. When the prediction shows that multiple future monitoring periods will have health scores below the adaptive threshold, the system issues an early warning, enabling maintenance personnel to arrange preventive maintenance before the actual occurrence of the fault. This realizes the transformation from "passive maintenance" to "proactive prevention" maintenance mode and effectively avoids unplanned power outages caused by sudden equipment failures. Example 9:

[0083] Based on Example 1, the visual interactive scheduling module includes: The fault equipment highlighting and positioning submodule is used to query the standard 3D model component corresponding to the fault equipment in the power supply BIM model based on the fault equipment number and equipment type output by the fault location and tracing module, highlight the standard 3D model component in the 3D visualization scene with a preset highlight color, and automatically focus the 3D scene view to the 3D spatial coordinate position of the fault equipment. The spare parts location retrieval submodule is used to send an inventory query request to the enterprise material management system through a standard API interface based on the spare parts material code of the faulty equipment. It receives the inventory quantity and three-dimensional spatial coordinates of the spare parts storage location of each work area's spare parts warehouse, filters out the spare parts storage locations with an inventory quantity greater than zero, calculates the spatial distance between each spare parts storage location and the faulty equipment, determines the spare parts storage location with the smallest spatial distance as the target spare parts location, and outputs the three-dimensional spatial coordinates of the spare parts storage location and the warehouse shelf information. The optimal maintenance personnel matching submodule is used to match the corresponding maintenance professional category according to the equipment type of the faulty equipment. Based on the maintenance professional category, the maintenance personnel's shift list and the maintenance personnel's skill list, it filters out the maintenance personnel who are currently on duty and have the corresponding professional skills, and sorts them in a priority order from high to low according to the preset skill level to determine the optimal maintenance personnel. The maintenance work order generation and push submodule is used to push maintenance work orders, which include the name of the faulty equipment, the three-dimensional spatial coordinates of the faulty equipment, the defect level, the optimal maintenance strategy, the coordinates of the spare parts storage location, the spare parts warehouse shelf information, and the contact information of the best maintenance personnel, to the handheld maintenance terminal of the railway power supply equipment of the best maintenance personnel via the railway private network using the HTTPS protocol.

[0084] In this embodiment, the preset highlight color is red, which is used to highlight faulty equipment in the 3D visualization scene of the power supply BIM model, distinguishing it from the default color of other normal equipment.

[0085] In this embodiment, the spare parts location retrieval submodule establishes a data connection with the enterprise material management system through a standard RESTful API interface. The inventory query request is in GET format, the URL path includes the spare parts material code parameter, and the request header includes an API authentication token. The response data returned by the enterprise material management system is in JSON format, containing the warehouse number, inventory quantity, and three-dimensional spatial coordinates of the spare parts storage location. axis, axis, (Axis values) and warehouse shelf information.

[0086] In this embodiment, the spatial distance between the spare parts storage location and the faulty equipment Calculated using the Euclidean distance formula: ; in The three-dimensional spatial coordinates of the faulty equipment. The three-dimensional spatial coordinates of the spare parts storage location.

[0087] In this embodiment, the professional matching rule for operation and maintenance personnel is as follows: substation equipment faults are matched with substation maintenance, feeder cable faults are matched with cable maintenance, and overhead contact line equipment faults are matched with overhead contact line maintenance. This rule is pre-stored in the system in the form of a mapping table from equipment type to professional category.

[0088] In this embodiment, the preset skill level priority order from high to low is as follows: senior technician has the highest priority, followed by technician, then senior worker, and finally intermediate worker. The best maintenance personnel are the on-duty maintenance personnel with the highest skill level in the screening results.

[0089] In this embodiment, the handheld maintenance terminal for railway power supply equipment is an industrial-grade rugged tablet computer that supports 4G / 5G railway private network communication. It has built-in lightweight BIM model viewing software and can load cached power supply BIM model 3D scene data offline. Maintenance work orders are encapsulated in JSON format and pushed to the handheld maintenance terminal via the railway private network through the HTTPS protocol. After receiving the work order, the terminal automatically pops up a notification reminder and stores it in the local work order list.

[0090] The beneficial effects of the above technologies are as follows: The fault equipment highlighting and positioning submodule of this invention realizes the visual positioning of fault equipment by using the three-dimensional spatial coordinates pre-stored in the power supply BIM model; the spare parts location retrieval submodule realizes the automatic matching of the optimal spare parts source by using the spare parts material code and spare parts storage location coordinates pre-stored in the power supply BIM model; the best maintenance personnel matching submodule matches the professional category based on the fault equipment type and combines the scheduling and skill information to screen the best personnel; and the maintenance work order generation and push submodule integrates all the above information into a complete work order and pushes it to the handheld terminal, forming a complete automated scheduling link from fault equipment positioning, spare parts inventory retrieval, maintenance personnel matching to work order generation and push. After receiving the work order, maintenance personnel can directly view the BIM three-dimensional model of the fault equipment on the terminal, navigate to the spare parts storage location, and understand the detailed maintenance strategy, realizing a seamless connection from fault diagnosis to maintenance execution. It eliminates the links of manual reporting, manual query of duty rosters, and manual filling of work orders between fault diagnosis and maintenance execution, fundamentally solving the time delay problem caused by the need for hierarchical reporting and manual scheduling after the manual inspection method discovers anomalies in the existing technology.

[0091] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of this invention and its equivalents, this invention also intends to include these modifications and variations.

Claims

1. A BIM-based railway power supply equipment operation and maintenance platform, characterized in that, include: The power supply BIM model building module is used to build power supply BIM models of each power supply unit along the railway line based on the standardized railway power supply BIM component family library, 3D spatial drawings of railway power supply equipment and electrical connection diagrams of railway power supply equipment. The power supply unit data acquisition and fusion module is used to collect time-series data of various types of substation equipment, various types of feeder cables and various types of contact network equipment of each power supply unit in each monitoring period, and fuse them to obtain the fused time-series status feature matrix of each power supply unit in each monitoring period. The power supply health status assessment module is used to obtain the comprehensive health score of each power supply unit in each monitoring period based on the fused time-series status feature matrix of each power supply unit in each monitoring period and the trained comprehensive health score model. By adaptively adjusting the comprehensive health evaluation threshold of the current monitoring period by the predicted pantograph status index when the train runs to the current power supply unit to be evaluated in each monitoring period, the adaptive comprehensive health evaluation threshold of each monitoring period is obtained. Based on the comprehensive health score of each power supply unit in each monitoring period and the adaptive comprehensive health evaluation threshold, the power supply health status evaluation result of each power supply unit in each monitoring period is output. The fault location and tracing module is used to trace the fault source and locate the faulty equipment based on the fused temporal state feature matrix of each power supply unit and the corresponding power supply BIM model when the power supply health status assessment result of each power supply unit in each monitoring period is lower than the adaptive comprehensive health evaluation threshold. It also makes a defect level judgment on the faulty equipment and provides the optimal maintenance strategy. The visual interactive scheduling module is used to visually identify and locate faulty equipment in the power supply BIM model based on the fault location and tracing results, retrieve the inventory location of the corresponding spare parts for the faulty equipment, match the best maintenance personnel, generate maintenance work orders and push them to the maintenance personnel's terminals.

2. The BIM-based railway power supply equipment operation and maintenance platform according to claim 1, characterized in that, The power supply BIM model building module includes: The component family library acquisition submodule is used to acquire standard 3D model components of each power supply unit, each substation equipment, each feeder cable and its accessories, and each contact network equipment in each power supply unit along the railway line, based on the standardized railway power supply BIM component family library. Each standard 3D model component contains the corresponding equipment type and the corresponding equipment model. The 3D spatial coordinate acquisition submodule is used to acquire the actual 3D spatial coordinates of each power supply unit, each feeder cable and its accessories, and each contact network equipment within the railway line based on the 3D spatial drawings of the railway power supply equipment, as well as the spatial location coordinates and corresponding quantities of spare parts for each power supply unit, each feeder cable and its accessories, and each contact network equipment. The electrical connection relationship acquisition submodule is used to acquire the electrical connection relationships between electrical components and equipment in each power supply unit along the railway line based on the electrical connection diagram of the railway power supply equipment. The BIM model assembly generation submodule is used to take each standard 3D model component as the basic component unit, place each basic component unit in the corresponding spatial position according to the 3D spatial coordinates of each electrical component and equipment, establish directed connection edges between each basic component unit according to the electrical connection relationship between each electrical component and equipment, and generate the power supply BIM model of each power supply unit along the railway line.

3. The BIM-based railway power supply equipment operation and maintenance platform according to claim 1, characterized in that, The power supply unit data acquisition and fusion module includes: The substation data acquisition submodule is used to collect data on the traction transformer oil temperature, dissolved acetylene concentration in the traction transformer oil, dissolved hydrogen concentration in the traction transformer oil, grounding current of the traction transformer core, partial discharge quantity inside the GIS, vibration spectrum of the GIS cabinet, SF6 gas concentration, and leakage current of the surge arrester in each power supply unit at each monitoring time during each monitoring period. It also arranges the same type of sensor data at all monitoring times within the same monitoring period in chronological order to form the time sequence data of each type of substation equipment in each monitoring period, and arranges the sensor data of all types of substation equipment at the same monitoring time within the same monitoring period in the order of the preset sensor channels to form the time sequence vector of the substation equipment at the corresponding monitoring time of the corresponding monitoring period. The feeder data acquisition submodule is used to collect sensor data on the distributed optical fiber temperature of the feeder cable sheath, the feeder cable grounding leakage current, and the feeder cable insulation loss coefficient of the feeder cable in each power supply unit at each monitoring time during each monitoring period. It also arranges the same type of sensor data at all monitoring times within the same monitoring period in chronological order to form the time sequence data of each type of feeder cable for each monitoring period. Finally, it arranges the sensor data of all types of feeder cables at the same monitoring time within the same monitoring period in the order of the preset sensor channels to form the feeder cable time sequence vector for the corresponding monitoring time of the corresponding monitoring period. The overhead contact line data acquisition submodule is used to collect data on the overhead contact line equipment in each power supply unit at each monitoring time during each monitoring period, including the tension of the overhead contact line droppers, the vibration amplitude of the overhead contact line brackets, the contact pressure of the overhead contact line pantograph, the leakage current of the overhead contact line insulators, the wear thickness of the overhead contact line conductors, and the tension of the overhead contact line catenary. It also arranges the same type of sensor data at all monitoring times within the same monitoring period in chronological order to form the chronological data of each type of overhead contact line equipment in each monitoring period. Finally, it arranges the sensor data of all types of overhead contact line equipment at the same monitoring time within the same monitoring period in the order of the preset sensor channels to form the chronological vector of the overhead contact line equipment at the corresponding monitoring time of the corresponding monitoring period. The time-series feature fusion submodule is used to concatenate the time-series vectors of substation equipment, feeder cable, and contact network equipment of the same power supply unit at the same monitoring time in the same monitoring period and at the same monitoring moment in the feature dimension to generate a fused time-series state feature row vector for each monitoring moment. The fused time-series state feature row vectors of all monitoring moments in the monitoring period are stacked in time sequence to obtain the fused time-series state feature matrix of each power supply unit in each monitoring period.

4. The BIM-based railway power supply equipment operation and maintenance platform according to claim 1, characterized in that, The power supply health status assessment module includes: The health score acquisition submodule is used to input the fused time-series status feature matrix of each power supply unit in each monitoring period into the trained comprehensive health score model, and output the comprehensive health score of each power supply unit through the trained comprehensive health score model. The pantograph status evaluation submodule is used to collect various types of operating status data of the train pantograph during each monitoring period, and calculate the current pantograph status index of the train pantograph during each monitoring period based on the remaining thickness of the pantograph carbon sliding plate, the pantograph-catenary contact pressure and the pantograph lifting action time in the various types of operating status data. The adaptive evaluation threshold determination submodule is used to calculate the comprehensive health evaluation threshold for each monitoring period based on the pantograph status index and the basic comprehensive health evaluation threshold of the train pantograph for each monitoring period. The status assessment result judgment submodule is used to calculate the difference between the comprehensive health score of each power supply unit and the adaptive comprehensive health evaluation threshold for each monitoring period. Based on the preset difference range into which the difference falls, the power supply health status assessment level corresponding to each power supply unit for each monitoring period is output.

5. A BIM-based railway power supply equipment operation and maintenance platform according to claim 4, characterized in that, The adaptive evaluation threshold determination submodule includes: The train status data acquisition unit is used to obtain the average speed of the train during its operation, which is used as the average speed of the train. It also obtains the distance from the current position of the train's pantograph to the corresponding power supply unit at the end of each monitoring period, which is used as the distance the train needs to travel in each monitoring period. The state index prediction execution unit is used to input the train's average driving speed, the train's distance to be traveled in each monitoring period, and the current pantograph state index of the train's pantograph in each monitoring period into the trained pantograph state index prediction model to obtain the predicted pantograph state index when the train runs to the current power supply unit to be evaluated. The adaptive adjustment coefficient calculation unit is used to calculate the adaptive adjustment coefficient for the corresponding monitoring period based on the predicted pantograph state index when the train runs to the current power supply unit to be evaluated. The adaptive evaluation threshold calculation unit is used to calculate the comprehensive health evaluation threshold for each monitoring period based on the basic comprehensive health evaluation threshold and the adaptive adjustment coefficient of the corresponding monitoring period.

6. A BIM-based railway power supply equipment operation and maintenance platform according to claim 1, characterized in that, The fault location and tracing module includes: The fault screening and judgment submodule is used to receive the power supply health status assessment results of each power supply unit in each monitoring period output by the power supply health status assessment module, and to screen out unhealthy power supply units whose comprehensive health score is lower than the adaptive comprehensive health evaluation threshold, mark them as power supply units to be repaired, and trigger the fault tracing process. The fault equipment tracing submodule is used to compare the standard fusion time sequence status feature matrix of each power supply unit in a healthy state with the fusion time sequence status feature matrix of the power supply unit to be repaired in the current monitoring period, determine the position of the sliding window with the largest deviation from the standard fusion time sequence status feature matrix in the current monitoring period, and back-map the sliding window position to the corresponding equipment type and sensing channel to determine the equipment number and equipment type of the faulty equipment. The fault equipment location submodule is used to query the three-dimensional spatial coordinates and power supply unit information of the fault equipment in the power supply BIM model based on the equipment number of the fault equipment, and determine the operation and maintenance work area corresponding to the fault equipment. The defect level determination submodule is used to calculate the defect level value of the faulty equipment based on the similarity value of the sliding window corresponding to the faulty equipment, and to determine the defect level of the faulty equipment based on the preset defect level range into which the defect level value falls. The maintenance strategy determination submodule is used to match the optimal maintenance strategy from a preset maintenance strategy knowledge base based on the equipment type and defect level of the faulty equipment.

7. A BIM-based railway power supply equipment operation and maintenance platform according to claim 6, characterized in that, The fault equipment tracing submodule includes: The traversal unit determination unit is used to determine the size of the traversal unit of the sliding window based on the evaluation level of the power supply unit corresponding to the comprehensive health status label of the power supply unit in the power supply health status evaluation results. The traversal unit is a rectangular window, and its row height and column width are determined according to the evaluation level. The similarity traversal execution unit is used to synchronously traverse the fusion time-series state feature matrix and the corresponding standard fusion time-series state feature matrix of the power supply unit to be inspected during the current monitoring period, using the rectangular sliding window determined by the traversal unit as the traversal unit. Each traversal extracts the current sub-matrix of the sliding window coverage area in the fusion time-series state feature matrix and the standard sub-matrix at the same position in the standard fusion time-series state feature matrix, calculates the similarity between the current sub-matrix and the standard sub-matrix, and records the similarity value corresponding to each sliding window position. The source identification and judgment unit is used to compare the similarity values ​​of all sliding window positions, locate the sliding window position with the smallest similarity value as the abnormal window position, and reverse map to the corresponding device type and specific sensing channel according to the matrix row index and column index corresponding to the abnormal window position in a preset order to determine the fault device number and device type corresponding to the abnormal window.

8. A BIM-based railway power supply equipment operation and maintenance platform according to claim 1, characterized in that, It also includes a health status trend prediction module, which includes: The time series acquisition submodule is used to acquire the comprehensive health score and adaptive comprehensive health evaluation threshold of each power supply unit for N consecutive historical monitoring periods before the current monitoring period, and to construct the time series sequence of the comprehensive health score and the time series sequence of the adaptive comprehensive health evaluation threshold of each power supply unit respectively, where N is the preset historical window length; The trend prediction submodule is used to input the time series sequence of the comprehensive health score of each power supply unit and the time series sequence of the adaptive comprehensive health evaluation threshold into the trained trend prediction model. The trend prediction model outputs the predicted comprehensive health score sequence and the predicted adaptive comprehensive health evaluation threshold sequence of each power supply unit for the next M monitoring periods, where M is the preset prediction step size. The fault prediction submodule is used to calculate the difference between the predicted comprehensive health score and the predicted adaptive comprehensive health evaluation threshold for each power supply unit in the next M monitoring periods. This difference is used as the prediction difference for each future monitoring period. Future monitoring periods with a prediction difference less than zero are selected as future fault risk periods. When the number of future fault risk periods is greater than or equal to... If so, a manual investigation and warning will be triggered.

9. A BIM-based railway power supply equipment operation and maintenance platform according to claim 1, characterized in that, The visual interactive scheduling module includes: The fault equipment highlighting and positioning submodule is used to query the standard 3D model component corresponding to the fault equipment in the power supply BIM model based on the fault equipment number and equipment type output by the fault location and tracing module, highlight the standard 3D model component in the 3D visualization scene with a preset highlight color, and automatically focus the 3D scene view to the 3D spatial coordinate position of the fault equipment. The spare parts location retrieval submodule is used to send an inventory query request to the enterprise material management system through a standard API interface based on the spare parts material code of the faulty equipment. It receives the inventory quantity and three-dimensional spatial coordinates of the spare parts storage location of each work area's spare parts warehouse, filters out the spare parts storage locations with an inventory quantity greater than zero, calculates the spatial distance between each spare parts storage location and the faulty equipment, determines the spare parts storage location with the smallest spatial distance as the target spare parts location, and outputs the three-dimensional spatial coordinates of the spare parts storage location and the warehouse shelf information. The optimal maintenance personnel matching submodule is used to match the corresponding maintenance professional category according to the equipment type of the faulty equipment. Based on the maintenance professional category, the maintenance personnel's shift list and the maintenance personnel's skill list, it filters out the maintenance personnel who are currently on duty and have the corresponding professional skills, and sorts them in a priority order from high to low according to the preset skill level to determine the optimal maintenance personnel. The maintenance work order generation and push submodule is used to push maintenance work orders, which include the name of the faulty equipment, the three-dimensional spatial coordinates of the faulty equipment, the defect level, the optimal maintenance strategy, the coordinates of the spare parts storage location, the spare parts warehouse shelf information, and the contact information of the best maintenance personnel, to the handheld maintenance terminal of the railway power supply equipment of the best maintenance personnel via the railway private network using the HTTPS protocol.

Images