A method for intelligent determination of a car coupler head coupler and fault protection decision

Abstract

The embodiment of the application provides a car coupler positive hook intelligent judgment and fault protection decision method, which is applied to the field of rail transit technology, and the method comprises the following steps: fusion analysis is carried out by collecting car coupler multi-source monitoring data to generate positive hook state and risk level judgment results; when an abnormal state is identified, a potential energy management alliance is created in the process of coasting, braking force is calculated based on dynamic operation parameters, and an alternative path is planned, and a cooperative protection strategy is generated; the strategy effect is verified and the state evolution trend is predicted during the execution process; a composite digital wear mark is generated by analyzing abnormal wear characteristics, and a replacement decision is made accordingly; acoustic fingerprint matching technology is used to select replacement parts, and a low-intensity directional magnetic field is used to guide precise assembly. The application realizes the whole-process closed-loop management of car coupler state intelligent perception, fault cooperative protection and precise maintenance, and improves the safety control level and operation efficiency of the car coupler.

Description

Technical Field

[0001] This application relates to the field of rail transit technology, and in particular to a method for intelligent determination and fault protection decision-making of coupler alignment. Background Technology

[0002] In railway transportation systems, the coupler is a key component connecting various carriages, and its normal condition directly affects the safety and efficiency of train operation. The correct coupling state of the coupler is the basis for ensuring reliable connection and smooth power transmission. However, in complex marshalling yard operations and train operation, the coupler is prone to miscoupling or hooking due to impact, wear, and other reasons. If not dealt with in time, it may lead to train separation and operation interruption, or even serious traffic accidents.

[0003] Existing technologies mainly rely on on-site personnel to judge the condition of couplers through traditional methods such as visual observation and hammering to listen for sounds. This method is highly subjective and easily affected by personnel experience, fatigue, and ambient light. It cannot effectively identify early signs of failure, and the discovery of safety hazards is delayed. When a coupler fails, the current approach is usually to isolate or handle it as an independent fault point, lacking linkage with adjacent couplers and the train control system. This results in the abnormal stress borne by the faulty coupler not being effectively transferred and redistributed, which may expand the scope of the fault and even affect adjacent cars, failing to form a system-level coordinated safety protection.

[0004] Therefore, there is a need in this field for a comprehensive solution that can achieve intelligent determination of coupler status, collaborative fault protection, closed-loop decision optimization, and integration of predictive maintenance, in order to overcome many shortcomings of existing technologies and improve the safety and intelligent operation and maintenance level of railway transportation. Summary of the Invention

[0005] The embodiments of this application provide a method for intelligent determination and fault protection decision-making of coupler alignment, realizing a coupler status management method with intelligent perception, accurate judgment and collaborative protection, so as to improve operational safety and maintenance efficiency. To achieve the above objectives, this application adopts the following technical solution:

[0006] A method for intelligent determination and fault protection decision-making of coupler alignment, the method comprising:

[0007] Collect multi-source monitoring data of the coupler, and generate a fusion judgment result of the coupler's positive hook status and fault risk level based on the multi-source monitoring data;

[0008] Based on the fusion judgment results, when a non-positive hook state or risk level exceeds a preset threshold is identified, during the shunting operation, based on the established network topology, the car where the faulty coupler is located and the adjacent cars that can communicate are selected, and a potential energy management alliance is created.

[0009] The dynamic operating parameters of each car in the potential energy management alliance are detected, the required braking force is calculated based on the dynamic operating parameters, and an alternative shunting path is planned. Based on the alternative shunting path, an alliance cooperative protection strategy including cooperative braking and path replanning is generated.

[0010] The alliance collaborative protection strategy is executed, and the speed and spacing parameters during the execution of the alliance collaborative protection strategy are collected. The speed and spacing parameters are compared with the expected parameters to generate a comparison deviation result. Based on the deviation result, the strategy verification result and the evolution prediction of the coupler state are generated.

[0011] Based on the strategy verification results and evolution prediction, the abnormal wear characteristics of the coupler tongue of the faulty car are extracted, and a composite digital wear mark is generated by combining the historical impact load data. A coupler tongue replacement instruction is generated according to the composite digital wear mark.

[0012] To verify the execution conditions of the hook tongue replacement command, the composite digital wear mark is compared with the standard features in the standard state sample library. When the comparison result exceeds the set deviation threshold, the acoustic fingerprint of the faulty hook tongue is collected and matched with the selected replacement hook tongue information.

[0013] Send instructions to the shared spare parts cabinet to provide the replacement hook information and generate an assembly path plan; during the replacement operation, assemble according to the assembly path plan, establish a low-intensity directional magnetic field, and guide the replacement hook to be calibrated to the standard coupling angle.

[0014] In some possible implementations, during the shunting operation, based on a given network topology, the car containing the faulty coupler and its communicable neighboring cars are selected, and a potential energy management alliance is created, including:

[0015] Based on the fusion judgment result, the train formation database is queried to determine the physical location relationship between the carriage where the faulty coupler is located and its adjacent carriages.

[0016] Detect the communication link status between the car containing the faulty coupler and each adjacent car;

[0017] Based on the physical location relationship and communication link status, a candidate set of carriages that can establish data interaction is selected from adjacent carriages;

[0018] A list of potential energy management alliance members is generated based on the candidate set;

[0019] Based on the member list, a data transmission channel is established between the member carriages to complete the creation of the potential energy management alliance.

[0020] In some possible implementations, the calculation of the required braking force based on the dynamic operating parameters includes:

[0021] The dynamic operating parameters of each car are obtained through the potential energy management alliance. These dynamic operating parameters include coupler difference data, buffer travel, and coupler force state.

[0022] Based on the dynamic operating parameters, a planning algorithm is used to calculate the safe speed threshold of each car under the current shunting conditions;

[0023] Based on the aforementioned safe speed threshold, combined with track gradient data and car spacing information, the braking force requirements of each car are determined.

[0024] Based on the braking force requirement, calculate the total braking force required by the potential energy management alliance.

[0025] In some possible implementations, the execution of the alliance-coordinated protection strategy includes:

[0026] Analyze the cooperative braking requirements and path replanning information in the alliance's cooperative protection strategy;

[0027] Based on the coordinated braking requirements obtained from the analysis, the execution sequence of braking commands for each carriage is determined;

[0028] According to the execution sequence, braking control commands are sent to each carriage within the potential energy management alliance;

[0029] Based on the braking control command and path replanning information, coordinated motion control is performed on each carriage within the potential energy management alliance, including:

[0030] Precisely adjust the braking pressure of the corresponding carriage according to the braking control command, and collect braking pressure data;

[0031] Based on the braking pressure data, the execution effect of the coordinated braking requirement is verified, the braking force distribution of each carriage is adjusted, and the corresponding carriage is controlled to adjust its running trajectory.

[0032] In some possible implementations, the generation of strategy verification results and coupler state evolution predictions based on the deviation results includes:

[0033] During the implementation of the alliance's collaborative protection strategy, data on the actual speed changes of each carriage were collected.

[0034] Collect data on the hook tongue swing angle and buffer compression.

[0035] The actual speed change data is compared with the expected speed change curve to calculate the speed tracking error value;

[0036] The hook tongue swing angle and the buffer compression amount are compared with the safe operating threshold to evaluate the response status of the mechanical components;

[0037] Based on the speed tracking error value and the response status of mechanical components, a multi-index fusion algorithm is used to generate strategy verification results including execution effect rating and risk level;

[0038] Based on the verification results of the strategy, the evolution trend of the coupler state is predicted.

[0039] In some possible implementations, generating the hook replacement instruction based on the composite digital wear mark includes:

[0040] Analyze the abnormal wear characteristics in the composite digital wear markers to determine the wear level and wear location distribution of the hook tongue;

[0041] Based on the wear level and wear location distribution, corresponding identification information is generated;

[0042] By integrating the wear level, wear location distribution, and identification information, a hook tongue replacement instruction including replacement priority is generated.

[0043] In some possible implementations, the method further includes:

[0044] When complex assembly conditions are identified, a 3D overlay view of the faulty coupler is generated, which integrates the working environment.

[0045] Establish a communication connection with a remote expert platform that integrates historical fault case data;

[0046] The three-dimensional overlay view and the collected fault feature data are sent to a remote expert platform to obtain a maintenance guidance plan adapted to the current working conditions.

[0047] Operation instructions are generated based on the aforementioned maintenance guidance plan;

[0048] Perform assembly operations based on the aforementioned operation guidelines, and collect tool posture data, assembly sequence information, and component alignment parameters during the operation.

[0049] The assembly quality is verified based on the collected assembly data, and the verification results are associated with the corresponding maintenance guidance plan and stored. The system judgment rules are then updated based on the associated stored data.

[0050] In some possible implementations, the acquisition of the acoustic fingerprint of the faulty hook tongue and matching it with selected replacement hook tongue information includes:

[0051] Collect the resonance spectrum of the faulty hook tongue under specific frequency excitation;

[0052] Extract a set of characteristic parameters characterizing the internal stress state of the material from the resonance spectrum;

[0053] Calculate the similarity value between the feature parameter set and the reference acoustic features of each spare part in the shared spare parts cabinet, and generate a matching result;

[0054] Based on the matching results, the target hook is determined from the replacement hooks that meet the preset matching threshold, and a selection result including the matching confidence level is generated.

[0055] In some possible implementations, the replacement operation involves assembling according to an assembly path plan, establishing a low-intensity directional magnetic field, and guiding the replacement hook tongue to be calibrated to a standard coupling angle, including:

[0056] Based on the assembly path planning, the coupler connection area is located, and a magnetic field space with a directional gradient is established.

[0057] The orientation and attitude of the replacement hook tongue are monitored by sensors placed in the magnetic field space.

[0058] The orientation and attitude are compared with the standard coupling angle to obtain attitude deviation information;

[0059] Based on the attitude deviation information, the spatial gradient distribution of the magnetic field is adjusted so that the replacement hook tongue moves toward the standard coupling angle under the action of magnetic force.

[0060] When the attitude deviation is less than the preset tolerance value, a calibration completion signal is output.

[0061] In some possible implementations, the method further includes:

[0062] Collect historical coupler inspection data, including positive coupler determination results, execution deviation data in strategy verification results, and composite digital wear markers;

[0063] Establish a mapping relationship between the composite digital wear markers and historical impact load data;

[0064] Based on the aforementioned correlation mapping analysis, the evolution trend of composite digital wear markers is analyzed, and wear rate acceleration features are extracted.

[0065] The wear rate acceleration characteristics are matched with the current impact load spectrum to identify the wear critical point under the current load mode;

[0066] The remaining service life prediction is corrected based on the identified wear threshold.

[0067] When the revised remaining service life prediction is lower than the preset safety threshold, a graded early warning message is generated, and a preventive maintenance instruction is generated.

[0068] As can be seen from the above technical solution, this application has the following beneficial effects:

[0069] 1. This method overcomes the limitations of traditional single-source coupler detection methods by constructing a multi-source monitoring and intelligent judgment system for couplers. It adopts a multi-modal sensing network tailored to coupler characteristics and processes the geometric, temperature, and mechanical data of the coupler through intelligent fusion algorithms to achieve accurate judgment of the coupler status. It can not only accurately identify the positive coupling status of the coupler, but also generate the coupler fault risk level based on multi-dimensional data analysis. This represents a technological leap from simple status detection to intelligent risk assessment. The multi-source information fusion system tailored to coupler characteristics significantly improves the reliability of coupler status detection and provides an accurate data foundation for coupler safety protection.

[0070] 2. This method addresses the problems of crude coupler fault protection and lack of scientific basis in maintenance decisions by establishing a coupler collaborative protection and predictive maintenance mechanism. It proposes a potential energy management alliance for the coupler system, achieving a smooth transition of coupler forces through multi-car cooperative braking. Based on digital characterization of coupler wear characteristics and acoustic matching technology, a complete coupler maintenance decision chain is constructed. By predicting the trend of coupler status, coupler maintenance decisions are transformed from experience-based to data-driven, achieving precise maintenance of coupler components and significantly improving the safety and service life of the coupler system. Attached Figure Description

[0071] The invention will now be further described with reference to the accompanying drawings.

[0072] Figure 1 A first flowchart provided for an embodiment of this application;

[0073] Figure 2 A second flowchart provided for embodiments of this application;

[0074] Figure 3 The third flowchart provided for the embodiments of this application. Detailed Implementation

[0075] The terms "first," "second," and "third," etc., used in this application specification, claims, and drawings are used to distinguish different objects, not to define a specific order.

[0076] In the embodiments of this application, the words "exemplary" or "for example" are used to indicate examples, illustrations, or descriptions. Any embodiment or design described as "exemplary" or "for example" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of the words "exemplary" or "for example" is intended to present the relevant concepts in a specific manner.

[0077] Research has revealed that traditional coupler management relies mainly on manual experience and single-point handling, which suffers from problems such as low detection accuracy, delayed response, and limited protective measures. Manual inspections are easily affected by subjective factors, and single sensors are not reliable enough under complex operating conditions. Fault handling often involves emergency braking, which can easily trigger a chain reaction. Maintenance decisions lack data support, making it difficult to achieve accurate maintenance, which seriously affects the safety and efficiency of railway transportation.

[0078] To address the aforementioned issues, this application provides a method for intelligent determination and fault protection decision-making regarding coupler alignment:

[0079] Example 1

[0080] Definitions:

[0081] Potential Energy Management Alliance: This refers to a temporary collaborative control group formed around the faulty car, along with its adjacent communicable cars, when an abnormal coupler condition is detected. Through data sharing and coordinated actions among members, this alliance achieves collaborative management of the kinetic and potential energy of the faulty car, avoiding the impact effects of braking a single car.

[0082] Composite digital wear marker: Based on multi-dimensional data, including abnormal wear characteristics and historical impact loads, a digital wear description is generated. This marker quantitatively characterizes the wear state of the hook tongue, providing data support for preventive maintenance.

[0083] Alliance Collaborative Protection Strategy: Potential Energy Management Alliance members have adopted a comprehensive protection plan, including coordinated braking and path adjustment, to deal with coupler failures. This strategy achieves safe mitigation of failures through the coordinated actions of multiple carriages.

[0084] Acoustic fingerprinting: A set of acoustic signal features that characterize the material state and structural features of a component by analyzing its resonance characteristics under specific frequency excitation. This method can detect internal defects and stress states of the component.

[0085] Low-intensity directional magnetic field: A technology for generating electromagnetic fields through precise control, which can form a magnetic field distribution with a directional gradient in a specific spatial region to guide metal parts to achieve precise positioning.

[0086] like Figure 1 As shown, specifically:

[0087] Intelligent Judgment and Alliance Creation Based on Multi-Source Monitoring

[0088] In the shunting area of ​​the railway marshalling yard, intelligent monitoring stations for coupler status are deployed. These stations employ a distributed sensor layout to ensure comprehensive coverage of passing carriages. When a carriage passes through the monitoring area at a specific speed, the system initiates the following detection process:

[0089] Step S101: Multi-source data acquisition and fusion determination

[0090] The coupler's overall status data is collected by a multimodal sensor array deployed at the monitoring station.

[0091] The visual inspection unit uses a high-resolution 3D linear array camera to collect stereo point cloud data of the coupler. The camera is arranged at a specific angle and spacing to ensure that it can capture the 3D geometric features of key parts of the coupler. The collected point cloud data is processed by a point cloud registration algorithm to reconstruct the 3D geometric model of the coupler. This model can accurately reflect the spatial positional relationship of each component of the coupler, including the opening and closing state of the hook tongue and the alignment of the hook body.

[0092] The thermal imaging unit uses a high-performance infrared focal plane array sensor to acquire the temperature distribution map of the coupler surface. The sensor has high temperature and spatial resolution and can detect subtle temperature anomalies. By analyzing the temperature distribution characteristics, especially the abnormally high temperature area, it can identify whether there is abnormal friction or overload. The temperature data analysis also takes into account the influence of ambient temperature and improves the detection accuracy through temperature compensation algorithm.

[0093] The mechanical sensing unit monitors the stress state of the hook tongue through a distributed strain gauge network. The strain gauges are arranged in a specific topology at key parts of the hook tongue to form a complete stress monitoring network. This network can monitor the stress distribution and changing trend in real time and detect abnormal stress states. The strain data adopts a dynamic sampling strategy, which automatically increases the sampling frequency when an anomaly is detected to ensure the integrity of the data.

[0094] The data processing unit adopts a multi-source data fusion algorithm based on deep belief networks. Through multi-layer nonlinear transformation, it extracts the spatial features of visual point clouds, the temperature distribution features of thermal imaging, and the mechanical features of strain gauges. The network design takes into account the characteristics of different sensor data and uses a dedicated feature extraction sub-network to process different types of data.

[0095] During the feature fusion stage, an attention mechanism is used to dynamically adjust the importance weights of different features. For example, when the visibility is low, the weight of visual features is automatically reduced while the weights of thermal imaging and mechanical features are increased. This fusion strategy ensures that reliable state determination results can be obtained under different working conditions.

[0096] The final layer of the network outputs two key indicators: the confidence level of the coupler status and the fault risk level. The confidence level of the coupler status is calculated based on the matching degree between the 3D coupler model and the standard model, while also considering temperature anomaly patterns and stress distribution characteristics. The fault risk level is derived by analyzing the correlation between historical fault data and the current status, using a probabilistic graphical model for reasoning. The risk level is divided into multiple levels, corresponding to different handling strategies.

[0097] Step S102: Potential Energy Management Alliance Creation

[0098] When the system determines that a certain carriage is in a non-coupling state and the risk level exceeds the preset threshold, the alliance creation process is initiated:

[0099] The system first queries the train formation database to obtain the precise location information of the faulty car within the train formation. The database stores complete train formation information, including parameters such as the type, weight, and braking performance of each car. Based on this information, the system determines the identifiers of the forward and backward adjacent cars of the faulty car.

[0100] The communication link status with adjacent carriages is detected through the onboard communication module. The detection process employs a multi-index evaluation method, including signal strength, communication latency, packet loss rate, and link stability. The system sets dynamic thresholds and automatically adjusts communication quality requirements based on the current operating environment. For example, in areas with strong electromagnetic interference, the requirements for signal strength are appropriately reduced, while the requirements for link stability are increased.

[0101] Based on the physical location relationship and communication link status assessment results, the system selects candidate sets of carriages that can establish stable data interaction from adjacent carriages. The selection process adopts a multi-objective optimization algorithm, which considers multiple factors such as communication reliability, data processing capability, braking system status and current location relationship. The algorithm outputs a member combination scheme.

[0102] Based on the screening results, a list of potential energy management alliance members is generated. The list includes the carriage identification, role assignment and responsibility definition of each member. The system establishes a dedicated data transmission channel between member carriages. This channel adopts a time-division multiple access mechanism to ensure the real-time transmission of control commands and the synchronous acquisition of data. The channel also has a redundancy design, which automatically switches to the backup channel when the main channel fails.

[0103] Technical effect analysis:

[0104] The multi-source data fusion judgment mechanism constructs a multi-dimensional cognitive system for coupler status by leveraging the complementary advantages of heterogeneous sensors. Visual sensors provide precise geometric information, thermal imaging sensors reveal invisible temperature anomalies, and mechanical sensors reflect the internal stress state. This multi-modal perception method ensures the accuracy and reliability of status judgment. Especially under complex working conditions, when the performance of one type of sensor deteriorates, other sensors can still provide reliable data.

[0105] The creation mechanism of the potential energy management alliance establishes a reliable collaborative control infrastructure through rigorous communication quality assessment and member selection. Traditional single-car braking methods often produce impact effects and even cause secondary accidents due to a lack of coordination. However, the alliance mechanism can achieve smooth dissipation of kinetic energy through coordination among members, thereby improving operational safety. The multi-objective optimization in the member selection process ensures the overall effectiveness of the alliance, enabling subsequent collaborative control to be executed efficiently.

[0106] Example 2

[0107] like Figure 2 As shown, specifically: Implementation and verification of collaborative protection strategies.

[0108] After the establishment of the potential energy management alliance, the system enters the collaborative protection phase. The core of this phase is to achieve safe control of faults through the coordinated actions of multiple carriages. The specific implementation process is as follows:

[0109] Step S201: Dynamic Parameter Acquisition and Strategy Generation

[0110] Dynamic operating parameters are collected through a distributed sensor network in the carriages of alliance members:

[0111] The spacing monitoring unit uses a high-precision laser rangefinder to monitor the relative distance changes between carriages in real time. The rangefinder adopts the phase measurement principle, which has high measurement accuracy and response speed. Through the Doppler effect compensation algorithm, the measurement error caused by the relative motion of the carriages is eliminated, further improving the measurement accuracy. At the same time, the three-dimensional motion attitude of the carriages is acquired through a miniature inertial measurement unit, including pitch angle, roll angle and yaw angle. The inertial unit adopts a multi-sensor fusion algorithm, combining data from accelerometers and gyroscopes, to output stable and reliable motion attitude information.

[0112] The mechanical condition monitoring unit monitors the buffer stroke using a high-precision displacement sensor. The sensor employs the magnetostrictive principle, featuring non-contact and high precision. Combined with data from a pressure sensor mounted on the buffer, the system calculates the buffer's energy absorption capacity and remaining buffer capacity. The coupler's stress state is monitored by a shear force sensor installed at the coupler neck. This sensor uses a strain gauge bridge structure, enabling the measurement of multi-dimensional stress conditions.

[0113] Based on these dynamic parameters, the system employs an advanced model predictive control algorithm for computational optimization. The algorithm first establishes an accurate multibody dynamics model based on the car mass distribution, track gradient characteristics, and environmental factors. This model considers factors such as coupling effects between cars, track friction characteristics, and air resistance, enabling it to accurately predict the system's dynamic response.

[0114] Based on the model, the algorithm employs a rolling optimization strategy to calculate the safe speed threshold for each car. The calculation of the safe speed threshold not only considers basic kinematic constraints but also pays special attention to the load-bearing capacity limitations of the faulty coupler. Through stress distribution analysis and fatigue damage assessment, it ensures that no additional damage is caused to the faulty component during braking.

[0115] Based on the difference between the safe speed threshold and the current actual speed, the system accurately calculates the required braking force. The braking force distribution adopts an optimization algorithm, which comprehensively considers the braking capacity, positional relationship and motion state of each carriage to achieve the optimal distribution of braking force. The distribution principle is to minimize the impact and wear caused by braking while ensuring safety.

[0116] The route planning module also generates alternative shunting routes. The planning of these routes takes into account the track conditions, switch status, and the operation of other vehicles. The planning algorithm uses a graph search method to find the optimal alternative route in the global route network. The route evaluation function takes into account multiple objectives such as route length, smoothness, safety, and avoidance of conflicts with other vehicles.

[0117] Step S202: Strategy Execution and Verification

[0118] The system executes the protection strategy according to the closed-loop control process to ensure the accurate implementation and timely adjustment of the strategy:

[0119] The control command parsing unit first analyzes the coordinated braking requirements. Based on the braking characteristics and positional relationships of each car, it determines the execution sequence of the braking commands. The timing arrangement follows the wave propagation principle to ensure the smooth transmission of the braking wave and avoid impact effects. The command timing also takes into account communication delays and actuator response times, and eliminates the influence of these factors through feedforward compensation.

[0120] The command distribution unit sends braking control commands to alliance members through a dedicated communication channel. The commands adopt a standardized format and include detailed parameters such as target braking force, execution time, ramp curve, and emergency plan. The system monitors braking pressure feedback and precisely adjusts the brake cylinder pressure through a high-performance proportional-integral-derivative controller. The controller parameters employ an adaptive adjustment strategy, automatically optimizing control performance based on changes in the characteristics of the braking system.

[0121] During execution, the data acquisition unit continuously collects speed and spacing parameters, and uses the Kalman filter algorithm for data fusion and state estimation. The filter model takes into account the dynamic characteristics of the system and the noise characteristics of the sensors, and can output smooth and reliable state estimates. The actual parameters are compared with the expected values ​​to generate a quantitative evaluation of the strategy execution effect.

[0122] Step S203: State Evolution Prediction

[0123] Based on historical and monitoring data on strategy execution performance, the system uses a long short-term memory network to establish a coupler state evolution model:

[0124] The network structure is specially designed to effectively handle long-term dependencies in time-series data. Input features include real-time status parameters of the coupler, environmental conditions, operating parameters, and historical maintenance records. The network selectively memorizes important status features and forgets irrelevant information through a multi-layer gating mechanism.

[0125] This network learns the state change patterns of couplers under similar working conditions to predict the development trends of key coupler parameters in the future. The predictions include indicators such as coupler tongue wear progression, changes in stress concentration zones, material fatigue accumulation, and the development of potential defects. The prediction results are output in probabilistic form, reflecting the uncertainty of the predictions.

[0126] The predictive model also establishes the correlation between environmental factors and operating parameters and the evolution of the coupler's condition. Through multi-factor coupling analysis, the model can assess the impact of different environmental conditions and operating strategies on the coupler's lifespan, providing support for optimization decisions. For example, the model can predict how the coupler's remaining service life will change under specific gradient and load conditions.

[0127] Technical effect analysis:

[0128] The combination of dynamic parameter acquisition and model predictive control enables a fundamental shift from passive response to proactive prevention. Traditional braking control, based on fixed thresholds and rules, is difficult to adapt to complex and ever-changing operating environments. In contrast, this method, through real-time dynamic modeling and rolling optimization, can generate the optimal control strategy according to specific operating conditions. This adaptive control approach significantly improves the system's flexibility and reliability.

[0129] The cooperative braking strategy effectively solves the impact effect problem caused by traditional single braking through precise timing control and force distribution. In practical applications, when a coupler malfunction is detected on a downhill section, the system achieves smooth dissipation of kinetic energy through the coordinated efforts of its members. The optimized distribution of braking force avoids local overload, reduces additional damage to faulty components, and significantly lowers the risk of decoupling.

[0130] The state evolution prediction function provides a scientific basis for maintenance decisions through deep learning models. This enables maintenance personnel to deploy maintenance resources in advance and optimize maintenance plans based on the prediction results. The multi-factor analysis capability of the prediction model helps identify key factors affecting the life of the coupler, thereby taking targeted measures to prevent major failures from occurring at the source.

[0131] Example 3

[0132] like Figure 3As shown, specifically: intelligent maintenance guidance and preventative maintenance.

[0133] Based on the preliminary assessment and execution results, the system enters the maintenance decision-making phase. The goal of this phase is to achieve precision and standardization in maintenance operations through intelligent technological means. The specific implementation process is as follows:

[0134] Step S301: Wear Analysis and Replacement Decision

[0135] The system comprehensively assesses the wear condition of the coupler tongue of a faulty train by analyzing multi-source data:

[0136] Geometric wear characteristics were obtained by comparing and analyzing 3D scanned point clouds with standard models. The scanning system uses structured light technology, which can quickly acquire high-precision point cloud data of the hook tongue surface. Through feature matching algorithms, parameters such as wear depth, wear profile and surface roughness are calculated. Principal component analysis is used to identify characteristic wear patterns and distinguish between normal wear and abnormal wear.

[0137] Internal defect characteristics are obtained through ultrasonic flaw detection signal analysis. The flaw detection system adopts phased array technology, which can realize multi-dimensional scanning detection. Signal features are extracted through wavelet transform to identify micro-cracks and material fatigue signs. The signal analysis also uses pattern recognition methods to distinguish different types of defect characteristics.

[0138] Historical load data is extracted from the onboard recording system, recording the impact loads borne by the coupler throughout its operating cycle. Data analysis includes load amplitude statistics, frequency analysis, and distribution characteristic extraction. Special attention is paid to the occurrence patterns of peak loads and cumulative damage effects.

[0139] These features are correlated with historical impact load data in the time and frequency domains to generate composite digital wear markers. The markers are in the form of multi-dimensional vectors to quantitatively characterize the overall wear state of the hook tongue. The vector dimensions include geometric dimension deviation, material property degradation, potential defect distribution, and historical damage accumulation. Each dimension is normalized to facilitate subsequent comparison and analysis.

[0140] The system performs similarity matching between composite digital wear markers and a standard condition sample library. The matching algorithm uses both Euclidean distance and cosine similarity metrics to evaluate the similarity of wear conditions from different perspectives. The sample library includes data on various typical wear conditions, and each sample has corresponding maintenance decisions and experience summaries.

[0141] When the overall similarity is lower than the safety threshold, the system generates a hook tongue replacement instruction. The instruction generation process also considers the wear trend and urgency. The replacement priority is determined based on the severity of wear. The priority evaluation adopts a multi-criteria decision-making method, which comprehensively considers factors such as safety risks, operational impact, and maintenance costs.

[0142] Step S302: Spare parts matching and assembly guidance

[0143] The system executes an intelligent replacement process to ensure the accuracy and efficiency of maintenance operations:

[0144] The acoustic matching unit first collects the acoustic fingerprint of the faulty hook tongue. The acquisition system uses a piezoelectric transducer to apply an excitation signal of a specific frequency. The signal frequency range has been optimized to effectively excite the characteristic vibration mode of the hook tongue. The resonance response is collected through a microphone array. The array arrangement has been acoustically optimized to accurately capture the spatial sound field characteristics.

[0145] Characteristic parameters such as resonance frequencies, quality factors, and damping coefficients were extracted through spectral analysis. The analysis process employed high-resolution spectral estimation methods to ensure the accuracy of the characteristic parameters. The feature extraction also considered the effects of temperature and environmental noise, and robustness was improved through compensation algorithms.

[0146] The spare parts management unit calculates the similarity between acoustic features and the baseline data of spare parts in the shared spare parts library. The matching algorithm uses a distance metric in a multidimensional feature space, and comprehensively considers the multidimensional similarity of frequency characteristics, mode shapes and material parameters. The algorithm outputs the matching score and confidence of each spare part and selects the best matching spare part.

[0147] The assembly guidance system generates detailed assembly path plans. The planning process uses motion planning algorithms, taking into account the motion constraints and obstacle avoidance requirements of the tools. The path includes the tool's motion trajectory, the order of component docking, and the tightening torque requirements. The planning results are displayed in a visual form to guide operators in completing assembly operations.

[0148] During the assembly process, a low-intensity directional magnetic field is generated by an electromagnetic array. The magnetic field is precisely controlled to form a magnetic field distribution with a directional gradient in a specific spatial area. The orientation and attitude of the replacement hook tongue are monitored in real time by sensors and compared with the standard coupling angle. Based on the attitude deviation information, the magnetic field gradient distribution is dynamically adjusted so that the replacement hook tongue can move smoothly to the standard coupling angle under the action of magnetic force.

[0149] Step S303: Preventive maintenance and update

[0150] The system establishes a continuous learning mechanism to constantly improve the scientific nature of maintenance decisions:

[0151] The data collection module comprehensively records all data from the entire maintenance process, including fault characteristic modes, load status correlations, maintenance operation parameters, and final maintenance results. The data is stored and labeled in a standardized format to ensure data integrity and consistency. The storage scheme takes into account the time-series characteristics of the data and supports the tracing and analysis of historical data.

[0152] The model update module updates the system's judgment rules and prediction models based on new maintenance data using online learning algorithms. The update process emphasizes incremental learning, incorporating new experiences while maintaining existing knowledge. In particular, the evaluation criteria for composite digital wear markers are adjusted according to actual maintenance results. The adjustment process employs a reinforcement learning framework, using a reward mechanism to guide the model's optimization direction.

[0153] The knowledge base maintenance module transforms successful maintenance cases into standard operating procedures (SOPs). This transformation process includes feature extraction, experience summarization, and procedure generation. Once validated, the new procedures are incorporated into the knowledge base, enriching the system's solution library. Cluster analysis identifies new failure modes, expanding the system's cognitive boundaries. Clustering algorithms can automatically discover hidden patterns in data, providing the system with new knowledge.

[0154] Technical effect analysis:

[0155] The comprehensive evaluation mechanism of composite digital wear markers establishes a scientific wear condition assessment system through cross-validation of multi-dimensional data. Traditional single-index assessments often only reflect one aspect of wear and cannot comprehensively assess the actual condition of components. However, this method achieves a more accurate condition assessment by integrating multi-source information such as geometry, materials, and loads. The assessment results not only reflect the current wear condition but also predict future development trends, providing strong support for preventive maintenance.

[0156] Acoustic fingerprint matching technology provides an objective basis for spare parts selection, ensuring the compatibility of replacement parts with the original system. Traditional spare parts selection mainly relies on model matching, which cannot take into account manufacturing tolerances and material differences. Acoustic matching, through the analysis of characteristic frequencies and vibration modes, can assess the degree of matching of the dynamic characteristics of spare parts, effectively avoiding secondary failures caused by component mismatch and significantly improving the repair success rate.

[0157] Magnetic field-assisted assembly technology has solved the technical challenge of precision assembly in confined spaces. Traditional assembly operations rely on the experience and skills of operators, making it difficult to guarantee quality stability. Magnetic field guidance, through non-contact precision control, achieves millimeter-level positioning accuracy, significantly improving assembly quality and efficiency. This technological advantage is particularly evident in situations with poor visibility or limited operating space.

[0158] The preventative maintenance and update mechanism enables the system to continuously evolve. Through the accumulation of maintenance experience, the system's decision-making accuracy gradually improves, forming a virtuous cycle of learning. This self-optimizing characteristic allows the system to adapt to new failure modes and changes in operating conditions, maintaining excellent performance over the long term.

[0159] The various technical features of this invention are closely logically linked and have multi-level synergistic relationships, forming a complete intelligent decision-making system:

[0160] Multi-source monitoring data provides the information foundation for fusion judgment, which determines the quality of all subsequent decisions. Visual sensors provide geometric information, thermal imaging sensors reveal temperature anomalies, and mechanical sensors reflect stress states. These heterogeneous data are fused through a deep belief network to form a comprehensive understanding of the coupler's condition. This multimodal perception approach ensures the accuracy of condition judgment and provides reliable input for subsequent decisions.

[0161] Accurate judgment is a prerequisite for creating an effective potential energy management alliance. Only based on reliable state identification can emergency mechanisms be correctly triggered. The alliance creation process itself is an optimization decision, ensuring the alliance's ability to perform coordinated control through communication quality assessment and member selection. The alliance's composition directly affects the effectiveness of subsequent coordinated control.

[0162] The establishment of the potential energy management alliance provides organizational support for the execution of the collaborative protection strategy. Data sharing and coordinated actions among alliance members make complex multi-vehicle collaborative control possible. The formulation of the collaborative protection strategy is based on real-time analysis of dynamic operating parameters, ensuring the scientific nature and adaptability of the strategy. The effectiveness of the strategy is then verified through a real-time monitoring system, forming a complete closed-loop control.

[0163] The data generated during strategy verification provides crucial input for state evolution prediction. This data not only reflects the current control effectiveness but also includes rich information about the system's dynamic response. Through analysis of the Long Short-Term Memory (LSTM) network, the system can learn the patterns of state evolution from this data and predict fault development trends. This predictive capability enables the system to take proactive measures.

[0164] Wear analysis based on prediction results guides maintenance decisions, ensuring the timeliness and accuracy of maintenance work. Wear analysis not only considers the current state but also predicts future trends, making maintenance decisions more scientific. New data generated during intelligent maintenance, including maintenance operation records and final maintenance results, in turn optimize the system's judgment rules and prediction models, forming a cycle of continuous improvement.

[0165] The foregoing has shown and described the basic principles, main features, and advantages of this application. Those skilled in the art should understand that this application is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of this application. Various changes and modifications can be made to this application without departing from the spirit and scope thereof, and all such changes and modifications fall within the scope of this application as claimed. The scope of protection of this application is defined by the appended claims and their equivalents.

Claims

1. A method for intelligent determination and fault protection decision-making of coupler alignment, characterized in that, The method includes: Collect multi-source monitoring data of the coupler, and generate a fusion judgment result of the coupler's positive hook status and fault risk level based on the multi-source monitoring data; Based on the fusion judgment results, when a non-positive hook state or risk level exceeds a preset threshold is identified, during the shunting operation, based on the established network topology, the car where the faulty coupler is located and the adjacent cars that can communicate are selected, and a potential energy management alliance is created. The dynamic operating parameters of each car in the potential energy management alliance are detected, the required braking force is calculated based on the dynamic operating parameters, and an alternative shunting path is planned. Based on the alternative shunting path, an alliance cooperative protection strategy including cooperative braking and path replanning is generated. The alliance collaborative protection strategy is executed, and the speed and spacing parameters during the execution of the alliance collaborative protection strategy are collected. The speed and spacing parameters are compared with the expected parameters to generate a comparison deviation result. Based on the deviation result, the strategy verification result and the evolution prediction of the coupler state are generated. Based on strategy verification results and evolution prediction, abnormal wear characteristics of the coupler tongue of the faulty car are extracted, and a composite digital wear mark is generated by combining historical impact load data. A coupler tongue replacement instruction is generated according to the composite digital wear mark. To verify the execution conditions of the hook tongue replacement command, the composite digital wear mark is compared with the standard features in the standard state sample library. When the comparison result exceeds the set deviation threshold, the acoustic fingerprint of the faulty hook tongue is collected and matched with the selected replacement hook tongue information. Send instructions to the shared spare parts cabinet to provide the replacement hook information and generate an assembly path plan; during the replacement operation, assemble according to the assembly path plan, establish a low-intensity directional magnetic field, and guide the replacement hook to be calibrated to the standard coupling angle.

2. The method according to claim 1, characterized in that, During the shunting operation, based on a predetermined network topology, the car containing the faulty coupler and its communicable neighboring cars are selected, and a potential energy management alliance is created, including: Based on the fusion judgment result, the train formation database is queried to determine the physical location relationship between the carriage where the faulty coupler is located and its adjacent carriages. Detect the communication link status between the car containing the faulty coupler and each adjacent car; Based on the physical location relationship and communication link status, a candidate set of carriages that can establish data interaction is selected from adjacent carriages; A list of potential energy management alliance members is generated based on the candidate set; Based on the member list, a data transmission channel is established between the member carriages to complete the creation of the potential energy management alliance.

3. The method according to claim 2, characterized in that, The calculation of the required braking force based on the dynamic operating parameters includes: The dynamic operating parameters of each car are obtained through the potential energy management alliance. These dynamic operating parameters include coupler difference data, buffer travel, and coupler force state. Based on the dynamic operating parameters, a planning algorithm is used to calculate the safe speed threshold of each car under the current shunting conditions; Based on the aforementioned safe speed threshold, combined with track gradient data and car spacing information, the braking force requirements of each car are determined. Based on the braking force requirement, calculate the total braking force required by the potential energy management alliance.

4. The method according to claim 1, characterized in that, The execution of the alliance-based collaborative protection strategy includes: Analyze the cooperative braking requirements and path replanning information in the alliance's cooperative protection strategy; Based on the coordinated braking requirements obtained from the analysis, the execution sequence of braking commands for each carriage is determined; According to the execution sequence, braking control commands are sent to each carriage within the potential energy management alliance; Based on the braking control command and path replanning information, coordinated motion control is performed on each carriage within the potential energy management alliance, including: Precisely adjust the braking pressure of the corresponding carriage according to the braking control command, and collect braking pressure data; Based on the braking pressure data, the execution effect of the coordinated braking requirement is verified, the braking force distribution of each carriage is adjusted, and the corresponding carriage is controlled to adjust its running trajectory.

5. The method according to claim 4, characterized in that, The generation of strategy verification results and coupler state evolution predictions based on the deviation results includes: During the implementation of the alliance's collaborative protection strategy, data on the actual speed changes of each carriage were collected. Collect data on the hook tongue swing angle and buffer compression. The actual speed change data is compared with the expected speed change curve to calculate the speed tracking error value; The hook tongue swing angle and the buffer compression amount are compared with the safe operating threshold to evaluate the response status of the mechanical components; Based on the speed tracking error value and the response status of mechanical components, a multi-index fusion algorithm is used to generate strategy verification results including execution effect rating and risk level; Based on the verification results of the strategy, the evolution trend of the coupler state is predicted.

6. The method according to claim 1, characterized in that, The step of generating a hook replacement instruction based on the composite digital wear mark includes: Analyze the abnormal wear characteristics in the composite digital wear markers to determine the wear level and wear location distribution of the hook tongue; Based on the wear level and wear location distribution, corresponding identification information is generated; By integrating the wear level, wear location distribution, and identification information, a hook tongue replacement instruction including replacement priority is generated.

7. The method according to claim 1, characterized in that, The method further includes: When complex assembly conditions are identified, a 3D overlay view of the faulty coupler is generated, which integrates the working environment. Establish a communication connection with a remote expert platform that integrates historical fault case data; The three-dimensional overlay view and the collected fault feature data are sent to a remote expert platform to obtain a maintenance guidance plan adapted to the current working conditions. Operation instructions are generated based on the aforementioned maintenance guidance plan; Perform assembly operations based on the aforementioned operation guidelines, and collect tool posture data, assembly sequence information, and component alignment parameters during the operation. The assembly quality is verified based on the collected assembly data, and the verification results are associated with the corresponding maintenance guidance plan and stored. The system judgment rules are then updated based on the associated stored data.

8. The method according to claim 1, characterized in that, The process of collecting the acoustic fingerprint of the faulty hook tongue and matching it with the selected replacement hook tongue information includes: Collect the resonance spectrum of the faulty hook tongue under specific frequency excitation; Extract a set of characteristic parameters characterizing the internal stress state of the material from the resonance spectrum; Calculate the similarity value between the feature parameter set and the reference acoustic features of each spare part in the shared spare parts cabinet, and generate a matching result; Based on the matching results, the target hook is determined from the replacement hooks that meet the preset matching threshold, and a selection result including the matching confidence level is generated.

9. The method according to claim 1, characterized in that, In the replacement operation, assembly is performed according to the assembly path plan, a low-intensity directional magnetic field is established to guide the replacement hook tongue to be calibrated to the standard coupling angle, including: Based on the assembly path planning, the coupler connection area is located, and a magnetic field space with a directional gradient is established. The orientation and attitude of the replacement hook tongue are monitored by sensors placed in the magnetic field space. The orientation and attitude are compared with the standard coupling angle to obtain attitude deviation information; Based on the attitude deviation information, the spatial gradient distribution of the magnetic field is adjusted so that the replacement hook tongue moves toward the standard coupling angle under the action of magnetic force. When the attitude deviation is less than the preset tolerance value, a calibration completion signal is output.

10. The method according to claim 1, characterized in that, The method further includes: Collect historical coupler inspection data, including positive coupler determination results, execution deviation data in strategy verification results, and composite digital wear markers; Establish a mapping relationship between the composite digital wear markers and historical impact load data; Based on the aforementioned correlation mapping analysis, the evolution trend of composite digital wear markers is analyzed, and wear rate acceleration features are extracted. The wear rate acceleration characteristics are matched with the current impact load spectrum to identify the wear critical point under the current load mode; The remaining service life prediction is corrected based on the identified wear threshold. When the revised remaining service life prediction is lower than the preset safety threshold, a graded early warning message is generated, and a preventive maintenance instruction is generated.

Images