A method and system for health monitoring of train traction systems

By combining real-time simulation models with 3D visualization models, maintenance animations are automatically matched and fault prediction is performed using offline multiphysics simulation. This solves the problem of manual reliance in train traction system monitoring and achieves efficient and visualized maintenance guidance and model self-calibration.

CN123114897APending Publication Date: 2026-11-17CRRC QINGDAO SIFANG ROLLING STOCK RESEARCH INSTITUTE CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611523383.X
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-09-29
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

The current health monitoring of train traction systems relies on manual inspections, which is inefficient and the quality is affected by the experience of maintenance personnel. It also lacks visual guidance and model self-calibration capabilities.

Method used

By combining real-time simulation models with 3D visualization models, fault codes are automatically matched with 3D maintenance animations. Combined with offline multiphysics simulation models, fault inversion and remaining life prediction are performed to achieve automated maintenance guidance and model calibration.

Benefits of technology

It improves the efficiency and standardization of fault handling, enhances the reliability of health monitoring results, enables fault prevention and real-time model updates, and provides intuitive maintenance guidance.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123114897A_ABST
    Figure CN123114897A_ABST
Patent Text Reader

Abstract

This invention relates to a method and system for health monitoring of a train traction system. The method includes the following steps: collecting operational status data of the train traction system and inputting it into a real-time simulation model for real-time simulation calculations, iteratively updating internal parameter information; comparing the updated internal parameter information values ​​with preset benchmark values, generating a unique fault code when the deviation exceeds a preset threshold; matching the fault code with a target 3D maintenance animation; marking faulty components in the 3D visualization model and automatically playing the animation to guide maintenance; receiving maintenance feedback information, parsing to obtain correction parameters, and updating the parameter iteration constraints of the real-time simulation model. This invention automatically transforms train fault early warning information into standardized maintenance guidance, effectively improving fault handling efficiency and the accuracy of long-term health monitoring.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention belongs to the field of rail vehicle technology, and in particular relates to a method and system for health monitoring of train traction systems. Background Technology

[0002] The train traction system is the core power source that provides power for train operation. The stability and reliability of its performance are directly related to the train's acceleration, braking, energy consumption control and overall operating efficiency. It is a core element for ensuring passenger safety and improving service quality.

[0003] Currently, the common practice for health monitoring of train traction systems is to collect operational status data of the train traction system through various sensors deployed on the system, input the operational status data into a pre-built real-time simulation model for real-time simulation calculations, and determine whether the train traction system has any abnormalities based on the calculation results.

[0004] When this type of health monitoring detects an anomaly, it can only notify the maintenance personnel of the abnormal component through alarm information. The specific maintenance work still depends on the maintenance personnel to consult the information and carry it out based on their personal experience, resulting in low maintenance efficiency and the maintenance quality being greatly affected by the experience level of the maintenance personnel. Summary of the Invention

[0005] The purpose of this invention is to solve one of the above-mentioned technical problems and to provide a method and system for monitoring the health of a train traction system.

[0006] To achieve the above objectives, the technical solution adopted by the present invention is as follows: A method for monitoring the health of a train traction system is applied to a train traction system health monitoring system. The system includes a real-time simulation model of the traction system, a three-dimensional visualization model of the traction system, and a mapping relationship database. The mapping relationship database stores a one-to-one mapping relationship between traction system fault codes and three-dimensional maintenance animations. The monitoring method includes the following steps: Collect operational status data of the train traction system; The running status data is input into the real-time simulation model for real-time simulation calculation, and the internal parameter information of the real-time simulation model is iteratively updated. The updated values ​​of internal parameter information are compared with preset baseline values. When the deviation between the two exceeds a preset threshold, a globally unique fault code is generated. The fault code includes at least the faulty component identifier and the fault type identifier. Based on the fault code retrieval mapping relationship database, the corresponding target 3D maintenance animation is obtained by matching; In the 3D visualization model, the faulty components indicated by the fault codes are marked, and the target 3D maintenance animation is played automatically to guide the maintenance work for the faulty components. The system receives maintenance feedback information after maintenance work is completed, parses the maintenance feedback information to obtain correction parameters, and updates the parameter iteration constraints of the real-time simulation model based on the correction parameters.

[0007] In some embodiments of the present invention, the system further includes an offline multiphysics simulation model of the traction system; the method further includes: Obtain maintenance condition data corresponding to maintenance operations; Input the correction parameters and maintenance condition data into the offline multiphysics simulation model, and iteratively update the coupling parameters of the offline multiphysics simulation model; Based on the updated offline multiphysics simulation model, fault inversion and remaining life prediction are performed on each component of the train traction system to obtain the health prediction results of each component. Based on the health prediction results, target components with potential hazards are identified and marked in a 3D visualization model.

[0008] In some embodiments of the present invention, the correction parameters are obtained by parsing maintenance feedback information, including: In response to the confirmation operation of maintenance completion, a maintenance message is generated as maintenance feedback information; Analyze the maintenance message and extract the component temperature threshold correction value and electrical parameter correction value from the maintenance message as correction parameters.

[0009] In some embodiments of the present invention, the parameter iteration constraints include parameter baseline constraints and safety threshold constraints; The parameter iteration constraints for updating the real-time simulation model based on corrected parameters include: Update the parameter baseline constraint terms based on the electrical parameter correction values; Update the safety threshold constraint based on the component temperature threshold correction value.

[0010] In some embodiments of the present invention, generating a globally unique fault code includes: The severity level of the fault is determined based on the degree of deviation between the updated value of the internal parameter information and the preset baseline value. A globally unique fault code is generated based on the faulty component identifier, fault type identifier, train identifier, and fault severity level.

[0011] In some embodiments of the present invention, marking the faulty component indicated by the fault code in the three-dimensional visualization model includes: Parse the fault code to obtain the fault severity level; The corresponding marking style is determined based on the severity level of the fault, and the faulty component is marked according to the determined marking style.

[0012] In some embodiments of the present invention, the three-dimensional visualization model is pre-generated through the following steps: According to the preset physical partitions of the train traction system, import the structural parameters of each component; the structural parameters include, but are not limited to, external dimensions, assembly coordinates, connection points between components and internal wiring routing. Based on structural parameters, a basic geometric model of each physical partition is generated, and disassembly and assembly constraints are configured for the detachable components. The disassembly and assembly constraints are used to record the disassembly and assembly sequence and connection position of the detachable components. Based on a pre-defined material parameter library, the system matches the corresponding component material with the material rendering parameters for each basic geometric model, and then renders each basic geometric model based on the matched material rendering parameters to obtain a 3D visualization model.

[0013] In some embodiments of the present invention, the method further includes: Bind disassembly interaction commands and parameter query commands to detachable components in 3D visualization models; In response to the disassembly interaction command for the target detachable component, the disassembly process of the target detachable component is displayed based on the disassembly and assembly sequence recorded by the disassembly and assembly constraint relationship. In response to a parameter query command for the target component, the system displays the real-time operating parameters of the target component output by the real-time simulation model.

[0014] In some embodiments of the present invention, the collection of operating status data of the train traction system includes: Sensing signals are collected by various sensors deployed in the train traction system, including but not limited to current, voltage, temperature, speed and vibration; The acquired sensor signals are converted from analog to digital to obtain external feature quantities; Edge calculation is performed based on external feature quantities, received control commands, and preset control parameters to obtain the calculated state data of the sensorless coverage area. External feature quantities and computational state data are preprocessed to obtain runtime state data; preprocessing includes, but is not limited to, time alignment, noise removal, and invalid data filtering.

[0015] In some embodiments of the present invention, the method further includes: Each time a parameter update of a real-time simulation model or an offline multiphysics simulation model is completed, the updated model parameters are stored and the corresponding parameter version is generated. In response to a model rollback command, the stored parameter version is retrieved, and the parameters of the corresponding model are restored to the retrieved parameter version.

[0016] In some embodiments of the present invention, the method further includes: The basic geometric models are lightweighted at multiple precision levels to obtain three-dimensional visualization models at multiple precision levels. Before marking the faulty component indicated by the fault code in the 3D visualization model, the following steps are also included: Obtain the current network bandwidth, and based on the current network bandwidth, select the target accuracy level 3D visualization model from multiple accuracy levels for loading.

[0017] Some embodiments of the present invention further provide a train traction system health monitoring system for implementing the above-described train traction system health monitoring method, including: A mapping database that stores a one-to-one mapping relationship between traction system fault codes and 3D maintenance animations; The data acquisition module is used to collect operating status data of the train traction system; The traction system real-time simulation model is used to perform real-time simulation calculations based on operating status data and iteratively update internal parameter information. The fault code generation module compares the updated values ​​of internal parameter information with preset baseline values. When the deviation between the two exceeds a preset threshold, a globally unique fault code is generated. The fault code includes at least a fault component identifier and a fault type identifier. The animation matching module is used to retrieve the mapping relationship database based on the fault code and match it to obtain the corresponding target 3D maintenance animation. A 3D visualization model of the traction system is used to mark the faulty components indicated by the fault codes and play the target 3D maintenance animation to guide the maintenance work for the faulty components. The model update module is used to receive maintenance feedback information after the maintenance work is completed, parse the maintenance feedback information to obtain correction parameters, and update the parameter iteration constraints of the real-time simulation model based on the correction parameters.

[0018] The beneficial effects of this invention are as follows: 1. This invention converts the internal parameter deviations of a real-time simulation model into fault codes, and automatically matches the target 3D maintenance animation based on the fault codes, so that maintenance work no longer relies on manual reference to materials and the personal experience of maintenance personnel, effectively improving the efficiency of fault handling and the standardization of operations. 2. This invention receives maintenance feedback information after maintenance work is completed, analyzes and obtains correction parameters, and updates the parameter iteration constraints of the real-time simulation model. This ensures that the model iteration always follows the actual state of the equipment, thereby effectively maintaining the output accuracy of the model and improving the reliability of health monitoring results. 3. This invention iteratively updates the coupling parameters by inputting the correction parameters and maintenance condition data into an offline multiphysics simulation model, and marks the target components with potential hazards in a three-dimensional visualization model, so that potential risks are presented in an intuitive way before the failure occurs, thus realizing proactive prevention before the failure occurs. 4. By importing structural parameters according to preset physical partitions to generate partition basic geometric models, and configuring disassembly and assembly constraints for detachable components to record the disassembly and assembly sequence and connection position, the three-dimensional visualization model can fully restore the internal structure, wiring and material characteristics of the traction equipment, providing a more realistic model carrier for fault location and maintenance guidance.

[0019] 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 pointed out in the description, claims and drawings. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0021] Figure 1 A flowchart of a method for health monitoring of a train traction system; Figure 2 This is a schematic diagram of the structure of a health monitoring method for a train traction system; Figure 3 This is a schematic diagram of the structure of the traction system three-dimensional visualization model provided in the embodiments of this application; Figure 4 This is a schematic diagram of the structure of the target component marking provided in the embodiments of this application; Figure 5 This is a structural schematic diagram of the three-dimensional inspection animation provided in the embodiments of this application. Detailed Implementation

[0022] To make the objectives, technical solutions, and advantages of this application clearer, the application is described and illustrated below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application. All other embodiments obtained by those skilled in the art based on the embodiments provided in this application without inventive effort are within the scope of protection of this application.

[0023] It should be noted that the terminology used herein is for the purpose of describing particular implementations only and is not intended to limit the exemplary implementations according to this application. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. Furthermore, it should be understood that the terms “comprising” and “having”, and any variations thereof, are intended to cover non-exclusive inclusion, for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such process, method, product, or apparatus.

[0024] Where there is no conflict, the embodiments and features in the embodiments of the present invention can be combined with each other.

[0025] The technical solution of the present invention will be described in detail below with reference to specific embodiments and accompanying drawings.

[0026] As attached Figure 1 - Appendix Figure 5 As shown in the illustrative embodiment of the train traction system health monitoring method of the present invention, the method is applied to the train traction system health monitoring system, which includes a real-time simulation model of the traction system, a three-dimensional visualization model of the traction system, and a mapping relationship database. The mapping relationship database stores a one-to-one mapping relationship between traction system fault codes and three-dimensional maintenance animations.

[0027] The monitoring method includes the following steps.

[0028] S1: Collects operational status data of the train traction system.

[0029] The train traction system refers to the complete power and electrical system that provides traction power for train operation, including but not limited to the traction control system host, traction rectifier, traction inverter, intermediate DC support circuit, power module (such as insulated gate bipolar transistor IGBT module), traction motor, high voltage circuit breaker, contactor, filter reactor, cooling fan, overvoltage absorption resistor and other electromechanical components.

[0030] In this embodiment, multi-source sensing acquisition technology is used. By deploying sensors at key measuring points of the train traction system, the physical quantities of each component are converted into transmittable electrical signals. The system then collects the electrical signals to obtain operating status data.

[0031] Specifically, a current sensor is placed on the grid side of the traction converter to collect the grid-side current; a voltage sensor is placed in the intermediate DC circuit to collect the DC voltage; a temperature sensor is placed on the heat sink of the power module to collect the module temperature; a temperature sensor and a speed encoder are placed on the traction motor to collect the motor stator temperature and motor speed; and a vibration sensor is placed on the equipment enclosure to collect vibration signals. All sensors can be directly connected to the traction control board via hardwires to achieve millisecond-level raw signal acquisition.

[0032] In one feasible implementation, the system performs periodic sampling according to a fixed sampling period to ensure data continuity.

[0033] In another feasible implementation, the system performs data acquisition upon receiving an external trigger command, which is used to reduce the amount of data under normal conditions.

[0034] The two implementation methods can be used in combination, namely, combining normal periodic data collection with encrypted data collection under abnormal operating conditions, thereby balancing the continuity of monitoring with the conservation of data transmission resources.

[0035] S2: Input the running status data into the real-time simulation model for real-time simulation calculation, and iteratively update the internal parameter information of the real-time simulation model.

[0036] Among them, the real-time simulation model of the traction system refers to a mathematical simulation model pre-constructed based on the electrical operation mechanism of the train traction system. It describes the electrical behavior of components such as traction rectifier, traction inverter, and traction motor using differential equations or equivalent circuit equations. It is deployed on the computing hardware on the vehicle side and is used to reproduce the operation process of the physical traction system in real time in virtual space.

[0037] Internal parameter information refers to parameters that characterize the electrical properties of components within a real-time simulation model but cannot be directly measured by sensors. Examples include the saturation voltage drop of an IGBT, the capacitance of the DC-side supporting capacitor, the stator and rotor resistance of a traction motor, and the reactance of a filter reactance. These parameters are used to depict the true electrical characteristics of the object being simulated by the model, and the accuracy of the internal parameter information determines the degree of agreement between the model output and the real system.

[0038] In this embodiment, the technologies employed are real-time digital simulation and parameter optimization. In one implementation, a particle swarm optimization algorithm is used for iterative updates. The algorithm treats each set of candidate internal parameter values ​​as a particle in the parameter space. Each particle constructs an objective function based on the error between the simulated external characteristic quantities (such as grid-side current, intermediate DC voltage, and motor speed) and the actually collected external characteristic quantities. The particle continuously adjusts its search direction and step size based on its own historical best position and the group's historical best position. After multiple iterations, the objective function converges, and the internal parameter values ​​corresponding to the particles at this point are the parameters that best match the actual device state.

[0039] Specifically, the on-board computing hardware adopts a collaborative architecture of main processor and coprocessor. The coprocessor receives operating status data through a high-speed bus and forwards it to the main processor. The main processor uses the operating status data as input to the real-time simulation model for real-time calculation to obtain the simulated external feature quantities output by the model. The main processor then compares the simulated external feature quantities with the actual external feature quantities in the operating status data to construct the objective function, runs an optimization algorithm to iteratively adjust the internal parameter information until the objective function meets the convergence condition or the number of iterations reaches the upper limit.

[0040] For example, taking the supporting capacitor of the intermediate DC circuit as an example, the capacitance value of the supporting capacitor will slowly decay after long-term operation. The real-time collected grid-side current, intermediate DC voltage and other operating status data are input into the real-time simulation model, and the model outputs the simulated intermediate DC voltage. The system constructs an objective function based on the error between the simulated voltage and the measured voltage, and iteratively adjusts the capacitance value parameters in the model through the particle swarm optimization algorithm, so that the simulated voltage gradually approaches the measured voltage. After the iteration converges, the capacitance value in the model is the current true equivalent capacitance value of the supporting capacitor.

[0041] S3: Compare the updated value of the internal parameter information with the preset baseline value. When the deviation between the two exceeds the preset threshold, generate a globally unique fault code. The fault code includes at least the fault component identifier and the fault type identifier.

[0042] Among them, the preset benchmark value refers to the benchmark value of the internal parameter information in the health state of the equipment. It is used as a reference standard to judge whether the internal parameters are abnormal. For example, it can be determined when the equipment leaves the factory or when the model is initially calibrated.

[0043] Fault coding refers to structured coding used to uniquely identify a fault event, which is used to transform fault information into a standardized index that can be directly retrieved by a database.

[0044] The fault type identifier is an identifier that indicates the type of fault, such as aging, over-temperature, or breakdown. Both are fields that must be included in the fault code.

[0045] It is understandable that component degradation often manifests first as a slow drift of its electrical parameters relative to a healthy baseline value. This drift does not cause obvious abnormalities in external operating conditions in the early stages, making it difficult to detect through manual inspection. In this embodiment, by comparing the updated values ​​of internal parameter information with preset baseline values, and generating a fault code containing a fault component identifier and a fault type identifier when the deviation exceeds a preset threshold, the early, hidden parameter drift is transformed into a structured fault event record.

[0046] For example, taking the aforementioned supporting capacitor as an example, after the model converges iteratively, it obtains the updated value of its capacitance. The updated value is compared with the preset reference value, i.e. the factory-calibrated capacitance value. If the deviation of the capacitance value exceeds the preset threshold, it is determined that the supporting capacitor has an aging-type abnormality and a fault code is generated. The fault component identifier points to the supporting capacitor, the fault type identifier indicates an aging-type fault, and the code also carries globally unique sequence information, so that this fault event can be uniquely identified throughout the entire line.

[0047] S4: Based on the fault code retrieval mapping relationship database, the corresponding target 3D maintenance animation is obtained.

[0048] Among them, the mapping relationship database refers to a pre-established database that stores the one-to-one mapping relationship between fault codes and 3D maintenance animations. It is used to provide standardized maintenance guidance resources that uniquely correspond to each type of fault.

[0049] like Figure 5 As shown, 3D maintenance animation refers to animation resources that demonstrate the standardized maintenance operation process for a specific fault in a 3D visualization manner. It is used to transform textual maintenance procedures into intuitive and viewable operation demonstrations.

[0050] Specifically, the system parses the fault code, extracts the faulty component identifier and fault type identifier, and uses the extracted identifiers as search criteria to query the mapping database. The mapping database returns the resource address of the 3D maintenance animation that uniquely corresponds to the search criteria, and the system retrieves the animation resource corresponding to that resource address as the target 3D maintenance animation.

[0051] In one feasible implementation, the mapping database is deployed on a ground-side server for centralized management and unified updates of animation resources.

[0052] In another feasible implementation, some commonly used mapping records of the mapping relationship database are cached on the vehicle side so that matching can still be completed when vehicle-to-ground communication is interrupted.

[0053] S5: Mark the faulty components indicated by the fault codes in the 3D visualization model and automatically play the target 3D maintenance animation to guide maintenance work on the faulty components.

[0054] Among them, the traction system 3D visualization model refers to a model that digitally expresses the shape, structure and spatial assembly relationship of each component of the train traction system in a 3D graphical manner. It is used to provide operation and maintenance personnel with virtual equipment that can be observed intuitively.

[0055] Specifically, based on the faulty component identifier in the fault code, the system locates the corresponding component model in the 3D visualization model and applies visual markers such as highlight masks to that component model. The system automatically pauses the current browsing interface of the 3D visualization model and plays the target 3D inspection animation in a pop-up window or full-screen mode.

[0056] For example, the content of the target 3D maintenance animation sequentially covers the complete operation process of fault confirmation, safe operation of equipment power failure, opening the outer casing, step-by-step disassembly of bolts, extraction of faulty parts, replacement of new parts, insulation testing, whole machine reset, and power-on verification.

[0057] S6: Receive maintenance feedback information after maintenance work is completed, parse the maintenance feedback information to obtain correction parameters, and update the parameter iteration constraints of the real-time simulation model based on the correction parameters.

[0058] Among them, maintenance feedback information refers to the information submitted to the system by the operation and maintenance side after the maintenance work is completed, which records the actual situation of this maintenance. It is used to transmit the real equipment status obtained at the maintenance site back to the system.

[0059] Specifically, after maintenance personnel complete their repair work, the system receives the repair feedback information they submit. This feedback information is then parsed to extract the recorded correction parameters. These correction parameters are then transmitted to the vehicle-mounted side via the vehicle-to-ground communication link and written into the parameter iteration constraints of the real-time simulation model. In the next iteration update, the real-time simulation model optimizes its internal parameters based on the updated parameter iteration constraints.

[0060] The embodiments of this application solve the problems of traditional monitoring relying on manual inspection, insufficient fault early warning, lack of visual guidance for maintenance, and inability of the model to self-calibrate. They realize a closed loop from real-time monitoring to visual maintenance and then to model calibration, thereby improving the visualization of traction system health monitoring.

[0061] In some embodiments of the present invention, the detection system further includes an offline multiphysics simulation model of the traction system.

[0062] The offline multiphysics simulation model for the traction system refers to a simulation model deployed on the ground that comprehensively considers the interactions between multiple physical fields, including mechanical, electrical, magnetic, and thermal fields. It is used to perform high-precision simulations of the traction system without real-time constraints. Compared to the on-board real-time simulation model, the offline multiphysics simulation model does not require millisecond-level solution speeds, thus allowing for the inclusion of more complete physical field coupling relationships.

[0063] The method also includes the following steps.

[0064] Obtain maintenance condition data corresponding to the maintenance operation.

[0065] Input the correction parameters and maintenance condition data into the offline multiphysics simulation model, and iteratively update the coupling parameters of the offline multiphysics simulation model.

[0066] Coupling parameters refer to parameters that describe the mutual influence between different physical fields such as mechanical, electrical, magnetic, and thermal fields. Examples include electrothermal coupling coefficient and electromagnetic coupling coefficient.

[0067] Correction parameters refer to the parameters obtained from maintenance feedback information and used to correct the model.

[0068] In this embodiment of the application, the equations of different physical fields are solved simultaneously by using coupling parameters.

[0069] The system uses the correction parameters and maintenance condition data as the input and reference target of the model. It uses an optimization algorithm (such as particle swarm optimization) to repeatedly adjust the value of the coupling parameters so that the error between the simulation output of the model and the real state reflected by the maintenance condition data converges. At this time, the value of the coupling parameters is the value that matches the real equipment.

[0070] For example, the system writes the corrected winding resistance value obtained from traction motor maintenance into the offline multiphysics simulation model, and uses the maintenance condition data as a reference. The model runs an electro-magnetic-thermal coupling simulation. The system compares the deviation between the simulated motor temperature rise and the actual measured temperature during maintenance, and iteratively adjusts the electro-thermal coupling coefficient of the model until the simulated temperature rise curve matches the measured data, thus completing the update of the coupling parameters.

[0071] Based on the updated offline multiphysics simulation model, fault inversion and remaining life prediction are performed on each component of the train traction system to obtain the health prediction results of each component.

[0072] Among them, fault inversion refers to using historical operating data to drive an offline model to reproduce the entire process of a component evolving from a normal state to a fault state, which is used to analyze the causes and evolution paths of faults.

[0073] Remaining life prediction refers to the estimation of the remaining service time of a component before it fails, based on its current condition and aging pattern.

[0074] Health prediction results refer to the conclusive information characterizing the health status of each component, obtained after performing fault inversion and remaining life prediction on each component. It is understandable that faults often undergo a long internal evolution process before evolving into visible anomalies, and component failures also exhibit a gradual aging pattern. In this embodiment, fault inversion and remaining lifetime prediction are performed based on an updated offline multiphysics simulation model. On the one hand, the evolution process of historical faults is fully reproduced to help locate the root cause of the fault. On the other hand, the current state of the components is extrapolated along the aging trajectory to obtain the health prediction results of each component.

[0075] Since the model's coupling parameters have just been calibrated using maintenance data, the above inferences are based on a more realistic picture of the equipment's condition, thus ensuring the accuracy of the predictions.

[0076] Based on the health prediction results, target components with potential hazards are identified and marked in a 3D visualization model.

[0077] Specifically, the system iterates through the health prediction results of each component, identifies components with remaining lifespan lower than expected or whose inversion conclusions indicate abnormalities as target components with potential risks, and searches for the corresponding component model in the 3D visualization model and applies visual markings.

[0078] In this embodiment, by acquiring maintenance condition data, inputting an offline multiphysics simulation model, iteratively updating coupling parameters, performing fault inversion and remaining life prediction, and marking potentially hazardous components in the 3D model, the problem of traditional digital twin models relying solely on historical driving data for iteration, failing to capture on-site maintenance conditions, and causing simulation deviations to expand over long-term operation is solved, thus achieving adaptive optimization and predictive maintenance of the offline model.

[0079] In some embodiments of the present invention, the method for obtaining correction parameters by parsing maintenance feedback information includes the following steps.

[0080] In response to the confirmation operation of maintenance completion, a maintenance message is generated as maintenance feedback information.

[0081] The confirmation operation for maintenance completion refers to the operation sent by the maintenance personnel to the system after completing the maintenance work to confirm the completion of the maintenance. For example, clicking the maintenance completion control in the 3D visualization interactive interface is used to trigger the system to officially record the maintenance information.

[0082] Specifically, the system provides a maintenance completion control in the 3D visualization interactive interface. After the maintenance personnel complete the maintenance work, they trigger this control. In response to this confirmation operation, the system assembles the relevant information of this maintenance into a maintenance message according to preset fields, and uses the maintenance message as maintenance feedback information.

[0083] In one feasible implementation, the maintenance message includes fields such as maintenance time, replacement parts information, correction threshold, and fault clearance status.

[0084] In another feasible implementation, some fields in the maintenance message are automatically filled in by the system based on the maintenance process, while the remaining fields are confirmed by the maintenance personnel.

[0085] Analyze the maintenance message and extract the component temperature threshold correction value and electrical parameter correction value from the maintenance message as correction parameters.

[0086] Among them, the component temperature threshold correction value refers to the correction amount for the component's temperature safety boundary, which is used to reflect the actual allowable temperature range change of the component after maintenance. The electrical parameter correction value refers to the correction amount for the component's electrical parameters (such as resistance, reactance, etc.), which is used to reflect the changes in the actual electrical characteristics after the component is replaced or repaired. Together, they constitute the correction parameters, which serve as the direct basis for subsequent model calibration.

[0087] In one feasible implementation, the system performs a reasonableness check on the modified value after extraction, for example, to determine whether it is within a reasonable numerical range, in order to prevent abnormal input values ​​from polluting the model.

[0088] In another feasible implementation, the system associates the extracted correction parameters with the fault code for this time, so that each model calibration can be traced back to the corresponding fault event.

[0089] In this embodiment, a maintenance message is generated by responding to the maintenance completion confirmation operation, and the temperature threshold correction value and electrical parameter correction value are parsed and extracted. This solves the problem that traditional maintenance records are difficult to automatically return and the model cannot absorb the field correction parameters, and realizes the structured transformation of maintenance data into model parameters.

[0090] In some embodiments of the present invention, the parameter iteration constraints include parameter baseline constraints and safety threshold constraints.

[0091] The method for updating the parameter iteration constraints of a real-time simulation model based on modified parameters includes the following steps.

[0092] Update the parameter baseline constraints based on the electrical parameter correction values.

[0093] Among them, the parameter baseline constraint term refers to the baseline term that constrains the iterative values ​​of the internal parameters. It is used to provide a parameter baseline for iterative updates, and the internal parameters obtained by iterative updates will be used as a reference to measure whether they are abnormal.

[0094] It is understandable that the actual electrical parameters change after a component is replaced. If the parameter benchmark on which the iteration is based is not updated, the model will exhibit systematic judgment bias. In this embodiment, by updating the parameter benchmark constraint terms based on the electrical parameter correction values, the actual electrical parameters confirmed by the maintenance are directly established as the new iteration benchmark. The update action is precisely applied to the terms in the constraint conditions corresponding to the electrical characteristics, without affecting other terms.

[0095] Update the safety threshold constraint based on the component temperature threshold correction value.

[0096] Among them, the safety threshold constraint term refers to the term in the parameter iteration constraint condition that constrains the fault judgment boundary. It is used to define the safety boundary on which the deviation judgment is based, that is, when it exceeds the boundary, it is judged as abnormal.

[0097] It is understandable that after a component undergoes maintenance, such as cleaning of the heat dissipation structure or replacement of heat dissipation components, its actual allowable safe temperature boundary may change. If the judgment boundary is not adjusted, the fault determination will become disconnected from the actual state of the component. In this embodiment, the actual safe boundary confirmed by maintenance is established as the new judgment boundary by updating the safety threshold constraint term based on the component temperature threshold correction value.

[0098] In this embodiment, the parameter baseline constraint terms are updated by electrical parameter correction values ​​and the safety threshold constraint terms are updated by temperature threshold correction values. This solves the problem that the model baseline and alarm threshold cannot be dynamically adjusted with on-site maintenance, and realizes closed-loop calibration of the constraints of the real-time simulation model.

[0099] In some embodiments of the present invention, the method for generating a globally unique fault code includes the following steps.

[0100] The severity level of the fault is determined based on the degree of deviation between the updated value of the internal parameter information and the preset baseline value.

[0101] The degree of deviation refers to the extent to which the updated value of the internal parameter information deviates from the preset benchmark value, and is used to quantify the severity of parameter drift.

[0102] Specifically, the system presets several progressively increasing deviation intervals, each corresponding to a fault severity level. The deviation between the updated value and the preset baseline value is calculated, and the interval into which the deviation falls is determined. The level corresponding to this interval is the fault severity level; the larger the deviation, the higher the corresponding fault severity level.

[0103] A globally unique fault code is generated based on the faulty component identifier, fault type identifier, train identifier, and fault severity level.

[0104] Among them, the train identifier refers to the information that uniquely identifies a train. It is used to indicate which train the fault occurred on, so that the fault code can still distinguish the source in a multi-train management scenario.

[0105] Specifically, the system concatenates and combines the fields according to the preset field arrangement rules to generate fault codes.

[0106] In one feasible implementation, the encoding also includes a fault occurrence time field to support fault tracing by time dimension.

[0107] In another feasible implementation, the encoding also includes a device subsystem identification field to support fault categorization by subsystem dimension.

[0108] In this embodiment, the severity level of the fault is determined by the degree of deviation, and a globally unique fault code is generated by combining the fault component identifier, fault type identifier, train identifier, and severity level. This solves the problem of incomplete fault information, inability to accurately retrieve and grade responses, and achieves the standardization and uniqueness of fault codes.

[0109] In some embodiments of the present invention, a method for marking faulty components indicated by fault codes in a three-dimensional visualization model includes the following steps.

[0110] Parse the fault code to obtain the fault severity level.

[0111] The corresponding marking style is determined based on the severity level of the fault, and the faulty component is marked according to the determined marking style.

[0112] The marking style refers to the specific way of visually marking components, such as different colors, transparency, or flashing effects, which are used to distinguish the severity of the fault by visual differences.

[0113] It is understandable that if all faults are marked in the same way, maintenance personnel cannot visually distinguish the urgency of their actions. In this embodiment, by determining a marking style based on the severity level of the fault and marking faulty components according to that style, the severity of the fault is transformed into an intuitive visual difference. Maintenance personnel can judge the urgency of their actions solely based on the marking style in the 3D visualization model. Specifically, minor warning faults are highlighted with a semi-transparent yellow light, while major hardware faults are highlighted with a flashing red light. Figure 4 As shown, the simulation parameters such as the name of the faulty component, the fault type, the real-time temperature, the loss, and the health status can be displayed simultaneously.

[0114] In some embodiments of the present invention, such as Figure 3 As shown, the 3D visualization model is pre-generated through the following steps.

[0115] According to the preset physical partitions of the train traction system, import the structural parameters of each component.

[0116] Among them, the preset physical partition refers to the area pre-divided according to the electrical structure and function of the train traction system. It is used to break down the complex traction equipment into several modeling units with clear boundaries.

[0117] Structural parameters refer to parameters that describe the geometry and assembly relationship of components. They include, but are not limited to, external dimensions (describing the geometric size and shape of the component), assembly coordinates (describing the installation position of the component in the equipment space), connection points between components (describing the mechanical connection positions between components), and internal wiring routing (describing the laying path of internal cables in the equipment). They are used to provide complete geometric and assembly basis for model generation.

[0118] Specifically, the system obtains the design drawing parameters of the equipment, classifies the parameters according to the preset physical partitions, with each partition corresponding to a type of electrical functional area, stores the classified parameters in a structured manner, and automatically maps them to the modeling units of the corresponding partitions.

[0119] In one feasible implementation, the preset physical partitions include a high-voltage main circuit area, a power module area, a motor stator and rotor area, and a heat dissipation structure area, so that the model corresponds one-to-one with the actual electrical structure of the traction equipment.

[0120] In another feasible implementation, the parameters of the detachable components and the parameters of the fixed housing are distinguished and marked separately during import, so as to provide a basis for subsequently distinguishing the detachable attributes of the components.

[0121] Based on structural parameters, a basic geometric model of each physical partition is generated, and disassembly and assembly constraints are configured for the detachable components.

[0122] Among them, the basic geometric model refers to the three-dimensional model generated according to the structural parameters, which describes the geometric shape and spatial position of each partition component, and is used to form the geometric body of the three-dimensional visualization model.

[0123] Assembly constraint relationships refer to information describing the rules for disassembling and assembling detachable components. They record the disassembly and assembly sequence and connection positions of detachable components (such as the number and position of bolts). They are used to provide a rule basis for subsequent disassembly demonstrations, so that the disassembly process conforms to the actual assembly process.

[0124] Based on a pre-defined material parameter library, the system matches the corresponding component material with the material rendering parameters for each basic geometric model, and then renders each basic geometric model based on the matched material rendering parameters to obtain a 3D visualization model.

[0125] The material parameter library refers to a database that pre-stores rendering parameters for various materials, which is used to provide rendering parameters corresponding to the real visual characteristics of parts made of different materials.

[0126] Material rendering parameters refer to parameters that control the visual effects of the surface of a part model (such as reflectivity, color, texture, and luminescence characteristics), used to make the model present a visual effect that matches the material of the real part.

[0127] In this embodiment, the material parameter library distinguishes between categories such as semiconductor materials, copper conductor materials, insulating materials, metal heat sink materials, and high-voltage insulator materials, in order to make each type of component present its own realistic texture.

[0128] For example, the system matches semiconductor material rendering parameters for IGBT power modules, copper material rendering parameters for connecting copper busbars, insulating material rendering parameters for insulating components, and aluminum alloy material rendering parameters for heat sink housings. After rendering, each component presents a visual effect consistent with the real device in the 3D visualization model, and high-temperature components present a gradient light effect, while high-voltage components present a warning light and shadow.

[0129] In this embodiment, by importing structural parameters in partitions, generating basic geometric models and configuring disassembly and assembly constraints, and matching material parameters for rendering, the problem that general 3D solutions cannot restore the internal electrical structure and maintenance disassembly relationship of traction equipment is solved, providing a highly realistic 3D carrier for fault location and standardized maintenance.

[0130] In some embodiments of the present invention, the method further includes the following steps.

[0131] Bind disassembly interaction commands and parameter query commands to detachable components in the 3D visualization model.

[0132] Among them, the disassembly interaction command refers to the interactive command used to trigger the disassembly display of the component, which allows maintenance personnel to actively view the disassembly process of the component through operation.

[0133] Parameter query commands are interactive commands used to query component operating parameters, enabling maintenance personnel to proactively obtain real-time status information of components.

[0134] In response to the disassembly interaction command for the target detachable component, the disassembly process of the target detachable component is displayed based on the disassembly and assembly sequence recorded by the disassembly and assembly constraint relationship.

[0135] The disassembly process refers to the process of disassembling components in accordance with the disassembly and assembly sequence recorded according to the disassembly and assembly constraints, and demonstrated in an animated or step-by-step interactive manner.

[0136] Specifically, the system receives disassembly interaction instructions for the target detachable component, reads the disassembly and assembly constraints of the component, obtains the recorded disassembly and assembly sequence, and displays each disassembly action in the 3D visualization model in sequence according to the disassembly and assembly sequence. For example, first show the shell being hidden, then show the connectors being removed, and finally show the component moving out along the detachment direction.

[0137] In response to a parameter query command for the target component, the system displays the real-time operating parameters of the target component output by the real-time simulation model.

[0138] It is understood that in this embodiment of the application, the real-time operating parameters of the target component are displayed by responding to parameter query commands, and the calculation results of the real-time simulation model are directly presented on the three-dimensional interface with the component as the object. Since the displayed parameters come from the real-time simulation model synchronized with the real equipment, maintenance personnel can understand the internal state of the component without disassembling the physical equipment, and the simulation value of the model is thus transformed into on-site usable state information.

[0139] In this embodiment, by binding disassembly interaction commands and parameter query commands, the disassembly process is displayed in the order of disassembly and assembly, and real-time operating parameters are displayed. This solves the problem that traditional two-dimensional interfaces cannot intuitively view the internal structure and real-time status, and realizes three-dimensional interactive monitoring and maintenance guidance.

[0140] In some embodiments of the present invention, the method for collecting operating status data of a train traction system includes the following steps.

[0141] Sensing signals are collected by various sensors deployed in the train traction system, including but not limited to current, voltage, temperature, speed, and vibration.

[0142] The acquired sensor signals are converted from analog to digital to obtain external characteristic quantities.

[0143] Analog-to-digital conversion (ADC) refers to the process of converting continuous analog electrical signals into discrete digital signals, which is used to make sensing signals into digital quantities that can be processed by computing devices.

[0144] External characteristic quantities refer to characteristic quantities that are directly converted from sensing signals and can be directly measured from outside the device (such as grid-side current, intermediate DC voltage, and motor speed), which are used as actual measurement references for model simulation.

[0145] In this embodiment, the technology used is signal conditioning and analog-to-digital conversion technology, which first filters and conditions the sensing signal, and then performs analog-to-digital conversion.

[0146] In one feasible implementation, filtering and conditioning employ a combination of hardware filtering and digital filtering to improve signal quality.

[0147] In another feasible implementation, the sampling rate of the analog-to-digital converter is set separately according to the signal type, which is used to balance conversion accuracy and processing burden.

[0148] Edge computing is performed based on external feature quantities, received control commands, and preset control parameters to obtain computational state data for sensorless coverage areas.

[0149] Among them, control commands refer to the commands issued by the train control system, such as traction commands and braking commands.

[0150] The preset control parameters refer to the control parameters of the traction control system itself, such as the drive pulse parameters of the power module, which are used to provide control-side information for equipment operation.

[0151] Edge computing refers to computing performed on vehicle-side computing units located close to the data source.

[0152] Understandably, due to cost, space, and reliability limitations, sensors cannot be installed in all parts of the equipment, thus creating monitoring blind spots. In this embodiment, edge computing is performed based on external characteristic quantities, control commands, and preset control parameters. By utilizing the inherent operating mechanism of the components, the calculated state data of unmeasurable parts can be deduced from measurable quantities. Therefore, the monitoring range is extended from the sensor coverage area to the entire internal domain of the equipment, and all calculations are completed locally on the vehicle side without increasing the vehicle-to-ground transmission burden.

[0153] The external characteristic quantities and computational state data are preprocessed to obtain the running state data.

[0154] Preprocessing refers to the processing of data before it enters the model. It includes, but is not limited to: time alignment (unifying data from different sources and with inconsistent sampling times to the same time base), noise removal (removing abnormal transition points in the signal), and invalid data filtering (removing missing or obviously distorted data segments). It is used to ensure that the data input into the model is real, synchronous, and effective.

[0155] In this embodiment, the preprocessing is completed on the vehicle side and then transmitted to the simulation computing unit via a high-speed bus to ensure low latency and integrity of data transmission.

[0156] In some embodiments of the present invention, the method further includes the following steps.

[0157] Each time a parameter update of a real-time simulation model or an offline multiphysics simulation model is completed, the updated model parameters are stored and a corresponding parameter version is generated.

[0158] Among them, parameter version refers to the versioned record of the model parameter status after each parameter update, which is used to ensure that each change in the model parameters can be independently identified and saved.

[0159] In response to a model rollback command, the stored parameter version is retrieved, and the parameters of the corresponding model are restored to the retrieved parameter version.

[0160] Among them, the model rollback instruction refers to the instruction to restore the model parameters to a historical version, which is used to trigger a recovery operation when the parameter update effect is not good.

[0161] It is understandable that model parameters change continuously with each maintenance and calibration. If only the latest state is retained, historical states cannot be traced, and any problems with updates cannot be recovered. In this embodiment, by storing the updated model parameters and generating corresponding parameter versions after each parameter update, the evolution process of model parameters is completely preserved in the form of a version sequence. The history of each calibration is traceable, and the management of model parameters thus has complete traceability.

[0162] In some embodiments of the present invention, the method further includes the following steps.

[0163] The basic geometric models are lightweighted with multiple precision levels to obtain three-dimensional visualization models with multiple precision levels.

[0164] Lightweight processing refers to reducing the amount of model data while ensuring model availability, thereby reducing the storage and transmission overhead of the model.

[0165] Multiple precision levels refer to generating multiple versions of the same model with different amounts of data and levels of precision, which are used to provide selectable model precision.

[0166] In this embodiment, the model data size is reduced through mesh simplification (reducing the number of meshes on the model surface), texture compression, and removal of internal details invisible from a distance. The greater the degree of simplification, the smaller the model data size and the lower the accuracy. The system applies different degrees of lightweighting to the basic geometric model to obtain 3D visualization models with multiple accuracy levels.

[0167] In some specific implementations, three levels of precision are generated: the first level is a simplified overall model, omitting internal details, used for long-distance overview and loading in weak network environments. The second level is a component-level model, retaining the core component structure, used for viewing individual components. The third level is a detailed model, fully preserving internal wiring and component details, used for inspection and disassembly. The three levels can be used in combination to cover all usage scenarios from overview to detailed inspection.

[0168] The following steps are included before marking the faulty component indicated by the fault code in the 3D visualization model.

[0169] Obtain the current network bandwidth, and based on the current network bandwidth, select the target accuracy level 3D visualization model from multiple accuracy levels for loading.

[0170] In this embodiment, the current network bandwidth is detected and compared with the data volume requirements of models at each accuracy level. The highest accuracy level that the current bandwidth can smoothly handle is selected as the target accuracy level, and the corresponding model is loaded. When the bandwidth changes, the system switches the accuracy level of the loaded model accordingly.

[0171] For example, when the train is running in a tunnel or the train-to-ground communication is in a weak network state, the system detects that the current network bandwidth is low and selects a low-precision 3D visualization model for loading, which is displayed smoothly. After the train returns to the depot, the network conditions improve, and a high-precision model is switched to be loaded, presenting complete internal structural details.

[0172] like Figure 2 - Figure 5 As shown, some embodiments of the present invention further provide a train traction system health monitoring system for implementing the above-described train traction system health monitoring method.

[0173] The monitoring system includes a mapping database, a data acquisition module, a real-time simulation model of the traction system, a fault code generation module, an animation matching module, a 3D visualization model of the traction system, and a model update module.

[0174] The mapping database stores a one-to-one mapping relationship between traction system fault codes and 3D maintenance animations.

[0175] It should be noted that the mapping relational database is the system's data storage component. Each record uniquely binds a fault code to a 3D maintenance animation, providing the data foundation for the automatic matching of fault codes to maintenance guidance resources. The database can be implemented using a relational database, with the fault code field as the retrieval key.

[0176] The data acquisition module is used to collect operational status data of the train traction system.

[0177] Specifically, the data acquisition module can adopt a combination of periodic acquisition and triggered acquisition: under normal conditions, data is acquired periodically to ensure continuity, and under abnormal operating conditions, data is acquired more intensively to increase the data density during critical periods.

[0178] The real-time simulation model of the traction system is used to perform real-time simulation calculations based on operating status data and iteratively update internal parameter information.

[0179] Specifically, the real-time simulation model can be deployed on computing hardware where the on-board main processor and coprocessor work together. The coprocessor receives data and forwards it to the main processor, which then runs an optimization algorithm to iterate internal parameters. The iteration can be continuous or triggered periodically.

[0180] The fault code generation module is used to compare the updated value of internal parameter information with the preset baseline value. When the deviation between the two exceeds the preset threshold, a globally unique fault code is generated. The fault code includes at least the fault component identifier and the fault type identifier.

[0181] Specifically, the fault code generation module can set thresholds for each parameter for judgment, or it can make a comprehensive judgment on multiple related parameters to suppress false judgments; the combination of the two methods can balance sensitivity and reliability.

[0182] The animation matching module is used to retrieve the mapping relationship database based on the fault code and match the corresponding target 3D maintenance animation.

[0183] Specifically, the animation matching module can parse fault codes to extract retrieval fields, query the mapping database, and retrieve the matched animation resources. It can be deployed on the ground and works in conjunction with the mapping database.

[0184] The 3D visualization model of the traction system is used to mark the faulty components indicated by the fault codes and play the target 3D maintenance animation to guide the maintenance work on the faulty components.

[0185] Specifically, the 3D visualization model can highlight faulty components and play animations covering the entire maintenance process in pop-up or full-screen mode, with subtitles and voice prompts. Its rendering output can be adapted to different display terminals.

[0186] The model update module is used to receive maintenance feedback information after the maintenance work is completed, parse the maintenance feedback information to obtain correction parameters, and update the parameter iteration constraints of the real-time simulation model based on the correction parameters.

[0187] Specifically, the model update module can parse maintenance feedback information to extract correction parameters, and send them to the vehicle side via the vehicle-to-ground communication link to update parameter iteration constraints; after the update, it can trigger immediate iterative verification, or it can iterate under typical working conditions.

[0188] In this embodiment, the data acquisition module, real-time simulation model, fault code generation module, animation matching module, 3D visualization model, and model update module work collaboratively with the support of a mapping relationship database. They respectively undertake the functions of data acquisition, real-time simulation, fault code generation, animation matching, visualization mark playback, and model backflow update. As a result, the system has fully realized a health monitoring closed loop from state perception to model self-updating in a component-based manner, solving the problem that existing systems can only issue alarms and cannot directly guide maintenance.

[0189] In some embodiments of the present invention, the system may adopt a layered deployment architecture, including a physical entity layer, an interactive perception layer, an in-vehicle virtual entity layer, a ground virtual entity layer, and a smart cloud platform layer.

[0190] The system comprises several layers: the physical entity layer, consisting of the electromechanical components of the train traction system and their associated sensing and acquisition hardware; the physical entity layer, including acquisition boards and edge computing units housed in the traction control cabinet, handles the processing and pre-processing of raw signals; the onboard virtual entity layer, deployed on onboard computing hardware, runs real-time simulation models; the ground virtual entity layer, deployed on ground servers, runs 3D visualization models and offline multiphysics simulation models; and the intelligent cloud platform layer, deployed on a distributed server cluster, handles the storage and management of global data. All layers are connected via a bus and vehicle-to-ground wireless communication link, forming a complete data pathway from data acquisition to cloud management.

[0191] In some embodiments of the present invention, the vehicle-side computing hardware of the system can adopt a collaborative architecture of a main processor and a coprocessor. The coprocessor receives preprocessed data and control commands from the acquisition stage via a high-speed bus and forwards them to the main processor via a high-speed interface. The main processor, as the runtime environment for the real-time simulation model, performs simulation calculations and parameter iterations. Data transmission between the vehicle-side and the ground side can employ a lightweight binary protocol to reduce the bandwidth consumption of the vehicle-to-ground wireless link.

[0192] In some embodiments of this invention, the intelligent cloud platform layer can deploy a distributed server cluster and incorporate a time-series database, a cache database, and a business database. Real-time simulation data, fault records, and maintenance operation records for all trains along the entire line are stored uniformly, supporting retrieval by train, time, component, and other dimensions. A global fault knowledge base and a 3D maintenance animation resource library are uniformly managed in the cloud. The platform can also achieve batch distribution of fault alarm messages from multiple trains and automatic generation of maintenance work orders, and store parameter files for model iteration and equipment lifecycle files, providing support for long-term data mining and analysis.

[0193] In some embodiments of the present invention, the system can also provide operation and maintenance training functions. The system calls an offline multiphysics simulation model to reproduce the complete evolution process of historical faults and automatically associates it with the corresponding 3D maintenance animation, generating training content that includes explanations of fault principles, fault evolution simulations, and disassembly and assembly animations. Operation and maintenance personnel can access the training content at any time through their terminals for learning. The system records the learning progress and forms a skills profile for the operation and maintenance personnel. Thus, historical fault cases are preserved as reusable training resources, and newly hired operation and maintenance personnel can learn fault handling without on-site supervision.

[0194] In some embodiments of this invention, data exchange between models within the system can be achieved through a unified data interaction mechanism. The output data of each model is uniformly accessed through the data interaction channel, and after format conversion, distributed to the models that require the data. For example, the output data of the real-time simulation model drives the state presentation of the 3D visualization model, the maintenance records generated by the 3D visualization model are used for parameter updates, and the prediction results of the offline multiphysics simulation model are sent to the 3D visualization model for hazard marking. Thus, the models are no longer isolated data silos, but rather a collaborative whole with updates and global linkage.

[0195] In some embodiments of this invention, the rendering output of the 3D visualization model can be adapted to various terminals. The system outputs rendering images of different specifications to adapt to the large display screens in the depot and the handheld terminals of maintenance personnel. The handheld terminals can also cache model resources and 3D inspection animations offline, which can still be accessed and used in environments without network access, thus adapting to operational scenarios such as power outage maintenance in the depot and emergency repairs in tunnels.

[0196] In some embodiments of the present invention, after generating a fault code, the system can also upload the fault code to the ground side via a vehicle-to-ground communication link, whereby the ground side performs retrieval of the mapping relationship database and animation matching. The system archives and stores records of fault events, maintenance operations, and model updates, forming an operation and maintenance data asset covering the entire lifecycle of the equipment. This data asset can be used for traction system design optimization, component life improvement, and operation and maintenance strategy iteration.

[0197] Finally, it should be noted that the various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The same or similar parts between the various embodiments can be referred to each other.

[0198] The above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them; although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications can still be made to the specific implementation of the present invention or equivalent substitutions can be made to some technical features without departing from the spirit of the technical solutions of the present invention, and all such modifications and substitutions should be covered within the scope of the technical solutions claimed in the present invention.

Claims

1. A method for health monitoring of a train traction system, characterized in that, An application is made to a train traction system health monitoring system. The system includes a real-time simulation model of the traction system, a three-dimensional visualization model of the traction system, and a mapping relationship database. The mapping relationship database stores a one-to-one mapping relationship between traction system fault codes and three-dimensional maintenance animations. The monitoring method includes the following steps: Collect operational status data of the train traction system; The running status data is input into the real-time simulation model for real-time simulation calculation, and the internal parameter information of the real-time simulation model is iteratively updated. Compare the updated value of the internal parameter information with the preset baseline value. When the deviation between the two exceeds a preset threshold, generate a globally unique fault code. The fault code includes at least a fault component identifier and a fault type identifier. Based on the fault code, the mapping relationship database is retrieved, and the corresponding target 3D maintenance animation is obtained. The faulty component indicated by the fault code is marked in the three-dimensional visualization model, and the target three-dimensional maintenance animation is played automatically to guide the maintenance work for the faulty component. The system receives maintenance feedback information after the maintenance operation is completed, parses the maintenance feedback information to obtain correction parameters, and updates the parameter iteration constraints of the real-time simulation model based on the correction parameters.

2. The train traction system health monitoring method according to claim 1, characterized in that, The system also includes an offline multiphysics simulation model of the traction system; the method further includes: Obtain the maintenance condition data corresponding to the maintenance operation; The correction parameters and the maintenance condition data are input into the offline multiphysics simulation model, and the coupling parameters of the offline multiphysics simulation model are iteratively updated. Based on the updated offline multiphysics simulation model, fault inversion and remaining life prediction are performed on each component of the train traction system to obtain the health prediction results of each component. Based on the health prediction results, target components with potential risks are identified and marked in the three-dimensional visualization model.

3. The train traction system health monitoring method according to claim 1, characterized in that, The process of parsing the maintenance feedback information to obtain correction parameters includes: In response to the confirmation operation of maintenance completion, a maintenance message is generated as maintenance feedback information; The maintenance message is parsed, and the component temperature threshold correction value and electrical parameter correction value in the maintenance message are extracted as the correction parameters.

4. The train traction system health monitoring method according to claim 3, characterized in that, The parameter iteration constraints include parameter baseline constraints and safety threshold constraints; The parameter iteration constraints for updating the real-time simulation model based on the corrected parameters include: The parameter baseline constraint terms are updated based on the electrical parameter correction values; The safety threshold constraint is updated based on the component temperature threshold correction value.

5. The method for monitoring the health of a train traction system according to claim 1, characterized in that, The generation of globally unique fault codes includes: The severity level of the fault is determined based on the degree of deviation between the updated value of the internal parameter information and the preset benchmark value. A globally unique fault code is generated based on the faulty component identifier, the fault type identifier, the train identifier, and the fault severity level.

6. The train traction system health monitoring method according to claim 5, characterized in that, The step of marking the faulty component indicated by the fault code in the three-dimensional visualization model includes: The fault code is parsed to obtain the fault severity level; Based on the severity level of the fault, a corresponding marking style is determined, and the faulty component is marked according to the determined marking style.

7. The train traction system health monitoring method according to claim 1, characterized in that, The 3D visualization model is pre-generated through the following steps: According to the preset physical partitions of the train traction system, import the structural parameters of each component; the structural parameters include, but are not limited to, external dimensions, assembly coordinates, connection points between components and internal wiring routing. Based on the structural parameters, a basic geometric model of each physical partition is generated, and disassembly and assembly constraints are configured for the detachable components therein; the disassembly and assembly constraints are used to record the disassembly and assembly sequence and connection position of the detachable components; Based on a preset material parameter library, material rendering parameters of corresponding component materials are matched for each basic geometric model, and each basic geometric model is rendered based on the matched material rendering parameters to obtain the three-dimensional visualization model.

8. The method for monitoring the health of a train traction system according to claim 7, characterized in that, The method further includes: Bind disassembly interaction commands and parameter query commands to the detachable components in the 3D visualization model; In response to a disassembly interaction command for a target detachable component, the disassembly process of the target detachable component is displayed based on the disassembly and assembly sequence recorded by the disassembly and assembly constraint relationship. In response to a parameter query command for a target component, the real-time operating parameters of the target component, output by the real-time simulation model, are displayed.

9. The method for monitoring the health of a train traction system according to claim 1, characterized in that, The collected train traction system operation status data includes: Sensing signals are collected by various sensors deployed in the train traction system, including but not limited to current, voltage, temperature, rotational speed, and vibration. The acquired sensing signals are converted from analog to digital to obtain external feature quantities; Edge calculation is performed based on the external feature quantities, the received control commands, and the preset control parameters to obtain the calculated state data of the sensorless coverage area. The external feature quantities and the computational state data are preprocessed to obtain the running state data; the preprocessing includes, but is not limited to, time alignment, noise removal, and invalid data filtering.

10. A train traction system health monitoring system, used to implement the train traction system health monitoring method according to any one of claims 1-9, characterized in that, include: A mapping database that stores a one-to-one mapping relationship between traction system fault codes and 3D maintenance animations; The data acquisition module is used to collect operating status data of the train traction system; A real-time simulation model of the traction system is used to perform real-time simulation calculations based on the operating status data and iteratively update internal parameter information. The fault code generation module is used to compare the updated value of the internal parameter information with the preset benchmark value. When the deviation between the two exceeds a preset threshold, a globally unique fault code is generated. The fault code includes at least a fault component identifier and a fault type identifier. Animation matching module is used to retrieve the mapping relationship database based on the fault code and match the corresponding target 3D maintenance animation; A 3D visualization model of the traction system is used to mark the faulty components indicated by the fault codes and play the target 3D maintenance animation to guide the maintenance work for the faulty components. The model update module is used to receive maintenance feedback information after the maintenance operation is completed, parse the maintenance feedback information to obtain correction parameters, and update the parameter iteration constraints of the real-time simulation model based on the correction parameters.