Training method, device and equipment of heavy haul train wheel-rail force monitoring model

Abstract

The invention discloses a training method, device and equipment for a heavy-load train wheel-rail force monitoring model. The training method comprises the steps that actual wheel-rail force and simulated wheel-rail force of a plurality of sets of rail inspection cars and heavy-load trains running on a preset rail are obtained; performing model training on each group of rail inspection vehicle simulation wheel-rail force and heavy-load train simulation wheel-rail force to obtain a simulation wheel-rail force mapping relation model of the rail inspection vehicle and the heavy-load train; and inputting the actual wheel-rail force of the track inspection vehicle and the actual wheel-rail force of the heavy-load train into the simulation wheel-rail force mapping relation model for neural network transfer learning to obtain a heavy-load train wheel-rail force monitoring model. On the basis, model training is carried out on multiple groups of simulation wheel-rail forces, so that a mapping relation between the heavy haul train and the track inspection vehicle is accurately determined, and then a simulation wheel-rail force mapping relation model is migrated from simulation data to reality in a transfer learning mode; therefore, the wheel-rail force detected by the heavy haul train wheel-rail force monitoring model is more accurate.

Description

Technical Field

[0001] The present application relates to the field of rail transit, and in particular to a training method, device and equipment for a heavy-load train wheel-rail force monitoring model. Background Art

[0002] With the extension of rail transit operating mileage and service life, the stability and safety of vehicle operation are becoming increasingly important. Considering the complexity of the vehicle operating environment, the dynamic interaction between wheels and tracks will produce potential risks of derailment and overturning. Among the many factors affecting railway safety, the measurement and analysis of wheel-rail forces plays a vital role. Wheel-rail forces directly reflect the interaction between vehicles and tracks and are the basis for operational safety assessment. Therefore, real-time detection and accurate prediction of wheel-rail forces are more realistic for the operational safety of vehicle / track systems.

[0003] At present, there are many methods in the industry for detecting the status of railway lines. Among them, a wheel-rail force measuring device and a measuring method proposed in the patent with publication number CN116620358A relate to the field of railway engineering technology. The wheel-rail force measuring device of the present invention includes a sensing component, a calibration component and a control component; the sensing component is arranged on the wheel-rail, and the sensing component has a lateral sensing unit and a vertical sensing unit; the calibration component includes a lateral calibration unit and a vertical calibration unit, the lateral calibration unit is used to make the lateral sensing unit generate a first sensing electrical signal, and the vertical calibration unit is used to make the vertical sensing unit generate a second sensing electrical signal; the control component is electrically connected to the lateral sensing unit and the vertical sensing unit respectively; wherein, in the calibration state, the control component receives the first sensing electrical signal and the second sensing electrical signal, and detects the safety state of the train when passing the wheel-rail according to the first sensing electrical signal and the second sensing electrical signal.

[0004] The above patent detects the wheel-rail force generated when a vehicle passes over a track by laying sensors at the wheel-rail. However, if the vehicle is to be detected in real time, sensors need to be installed at every location of the wheel-rail. This is undoubtedly unrealistic and costly, and thus it is impossible to achieve real-time detection of the vehicle's wheel-rail force. Summary of the invention

[0005] The embodiments of the present application provide a training method, device and equipment for a heavy-load train wheel-rail force monitoring model, which improves the real-time monitoring of vehicle wheel-rail force.

[0006] According to a first aspect of the present application, an embodiment of the present application provides a training method for a heavy-duty train wheel-rail force monitoring model, which may include:

[0007] Obtain the actual wheel-rail forces and simulated wheel-rail forces of multiple groups of track inspection vehicles and heavy-load trains running on preset tracks. The actual wheel-rail forces include: the actual wheel-rail forces of the track inspection vehicles and the actual wheel-rail forces of the heavy-load trains. The simulated wheel-rail forces include: the simulated wheel-rail forces of the track inspection vehicles and the simulated wheel-rail forces of the heavy-load trains.

[0008] Model training is performed on each set of track inspection vehicle simulated wheel-rail force and heavy-load train simulated wheel-rail force to obtain a simulated wheel-rail force mapping relationship model between the track inspection vehicle and the heavy-load train;

[0009] The actual wheel-rail force of the track inspection vehicle and the actual wheel-rail force of the heavy-haul train are input into the simulated wheel-rail force mapping relationship model for neural network transfer learning to obtain the heavy-haul train wheel-rail force monitoring model.

[0010] Optionally, the method further comprises:

[0011] Obtaining a track inspection vehicle structure model of a track inspection vehicle, a heavy-load train structure model of a heavy-load train, and road condition information of a preset track;

[0012] According to the track inspection vehicle structure model, the heavy-load train structure model and the road condition information, a track inspection wheel-rail force simulation model of the track inspection vehicle passing through the preset track and a heavy-load train wheel-rail force simulation model of the heavy-load train passing through the preset track are respectively constructed;

[0013] The track inspection vehicle simulation wheel-rail force and the heavy-load train simulation wheel-rail force are determined respectively by the track inspection vehicle simulation wheel-rail force model and the heavy-load train simulation wheel-rail force.

[0014] Optionally, the method further comprises:

[0015] Get a comprehensive map of the route of the preset track;

[0016] Analyze the comprehensive route map to obtain the road condition information of the preset track.

[0017] Optionally, the method further comprises:

[0018] Obtaining wheel-rail forces measured by the track inspection vehicle and the wheel-rail forces measured by the sensor when the track inspection vehicle passes through the preset track;

[0019] The wheel-rail force measured by the sensor is used as the standard value, and the wheel-rail force measured by the track inspection vehicle is calibrated to obtain the actual wheel-rail force.

[0020] Optionally, the wheel-rail force measured by the sensor is used as a standard value to calibrate the wheel-rail force measured by the track inspection vehicle to obtain the actual wheel-rail force of the track inspection vehicle, including:

[0021] The wheel-rail force measured by the track inspection vehicle and the wheel-rail force measured by the sensor are input into the BP neural network model for calculation to obtain the actual wheel-rail force.

[0022] Optionally, obtaining the wheel-rail force measured by the track inspection vehicle and the wheel-rail force measured by the sensor when the track inspection vehicle passes through the preset track includes:

[0023] The wheel-track force of the track inspection vehicle is obtained by using the force measuring wheel pair technology when the track inspection vehicle passes through the preset track;

[0024] The sensor wheel-rail force of the track inspection vehicle passing through the preset track is obtained through the ground sensor, and the ground sensor is connected to the preset track.

[0025] Optionally, model training is performed on each set of track inspection vehicle simulated wheel-rail force and heavy-load train simulated wheel-rail force to obtain a simulated wheel-rail force mapping relationship model of the track inspection vehicle and the heavy-load train, including:

[0026] The simulated wheel-rail force of the track inspection vehicle is used as the model input, and the simulated wheel-rail force of the heavy-load train is used as the training label to train the deep residual network model, and the simulated wheel-rail force mapping relationship model of the track inspection vehicle and the heavy-load train is obtained.

[0027] Optionally, the actual wheel-rail force of the track inspection vehicle and the actual wheel-rail force of the heavy-haul train are input into the simulated wheel-rail force mapping relationship model for neural network transfer learning to obtain a heavy-haul train wheel-rail force monitoring model, including:

[0028] The actual wheel-rail force of the track inspection vehicle is used as the model input, and the actual wheel-rail force of the heavy-load train is used as the training label to train the simulation wheel-rail force mapping relationship model, and the heavy-load train wheel-rail force monitoring model is obtained.

[0029] According to a second aspect of the present application, a training device for a heavy-load train wheel-rail force monitoring model is provided, which may include:

[0030] The first acquisition module is used to obtain the actual wheel-rail force and simulated wheel-rail force of multiple groups of track inspection vehicles and heavy-load trains running on preset tracks, wherein the actual wheel-rail force includes: the actual wheel-rail force of the track inspection vehicle and the actual wheel-rail force of the heavy-load train, and the simulated wheel-rail force includes: the simulated wheel-rail force of the track inspection vehicle and the simulated wheel-rail force of the heavy-load train;

[0031] A training module, used for performing model training on each group of simulated wheel-rail forces of the track inspection vehicle and the simulated wheel-rail forces of the heavy-load train, to obtain a mapping relationship model of the simulated wheel-rail forces of the track inspection vehicle and the heavy-load train;

[0032] The transfer learning module is used to input the actual wheel-rail force of the track inspection vehicle and the actual wheel-rail force of the heavy-load train into the simulated wheel-rail force mapping relationship model to perform neural network transfer learning to obtain a heavy-load train wheel-rail force monitoring model.

[0033] According to a third aspect of the present application, an embodiment of the present application provides a training device for a heavy-load train wheel-rail force monitoring model, which includes: a processor and a memory storing computer program instructions;

[0034] When the processor executes the computer program instructions, it implements the training method of the wheel-rail force monitoring model of a heavy-load train as described in any one of the first aspects.

[0035] The technical solution provided by the embodiments of the present application brings at least the following beneficial effects:

[0036] The embodiments of the present application provide a training method, device and equipment for a heavy-load train wheel-rail force monitoring model, which obtains actual wheel-rail forces and simulated wheel-rail forces of multiple groups of track inspection vehicles and heavy-load trains traveling on preset tracks; performs model training on each group of simulated wheel-rail forces of the track inspection vehicle and the simulated wheel-rail forces of the heavy-load train to obtain a simulated wheel-rail force mapping relationship model of the track inspection vehicle and the heavy-load train; inputs the actual wheel-rail forces of the track inspection vehicle and the actual wheel-rail forces of the heavy-load train into the simulated wheel-rail force mapping relationship model for neural network transfer learning to obtain a heavy-load train wheel-rail force monitoring model. Based on this, by training models for multiple sets of track inspection vehicle simulated wheel-rail forces and heavy-load train simulated wheel-rail forces, the mapping relationship between the heavy-load train and the track inspection vehicle can be accurately determined when the wheel-rail forces are obtained by simulation calculation. Then, through transfer learning, the simulated wheel-rail force mapping relationship model is transferred from the simulation data to reality, so that the wheel-rail forces detected by the heavy-load train wheel-rail force monitoring model are more accurate, which improves the safety and reliability of railway transportation and reduces the probability of accidents. At the same time, the heavy-load train wheel-rail force monitoring model can also be used to monitor the wheel-rail forces generated by heavy-load trains when they are running on tracks for a long time, without the need to lay a large number of sensors for detection. This can provide a scientific basis for the maintenance and management of railway infrastructure, extend the service life of the line, and save detection costs.

[0037] It should be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] The drawings herein are incorporated into the specification and constitute a part of the specification, illustrate embodiments consistent with the present application, and together with the specification are used to explain the principles of the present application, and do not constitute an improper limitation on the present application.

[0039] Figure 1 is a flow chart of a training method for a heavy-load train wheel-rail force monitoring model according to an exemplary embodiment;

[0040] Figure 2 is another flow chart of a training method for a heavy-load train wheel-rail force monitoring model according to an exemplary embodiment;

[0041] Figure 3 is a structural schematic diagram of a training device for a heavy-load train wheel-rail force monitoring model according to an exemplary embodiment;

[0042] Figure 4 It is a structural block diagram of a training device for a heavy-load train wheel-rail force monitoring model according to an exemplary embodiment. DETAILED DESCRIPTION

[0043] In order to more clearly understand the above-mentioned objectives, features and advantages of the present disclosure, the scheme of the present disclosure will be further described below. It should be noted that the embodiments of the present disclosure and the features in the embodiments can be combined with each other without conflict.

[0044] In the following description, many specific details are set forth to facilitate a full understanding of the present disclosure, but the present disclosure may also be implemented in other ways different from those described herein; it is obvious that the embodiments in the specification are only part of the embodiments of the present disclosure, rather than all of the embodiments.

[0045] As described in the background technology, the wheel-rail force generated when a vehicle passes over a track is detected by laying sensors at the wheel-rail. However, if the vehicle is to be detected in real time, sensors need to be installed at every location of the wheel-rail. This is undoubtedly unrealistic and costly, and thus it is impossible to achieve real-time detection of the vehicle's wheel-rail force.

[0046] Based on this, the present application provides a training method, device and equipment for a heavy-duty train wheel-rail force monitoring model. The following first introduces the training method for a heavy-duty train wheel-rail force monitoring model provided in an embodiment of the present application.

[0047] Embodiment 1;

[0048] Figure 1 A schematic diagram of a process of training a heavy-load train wheel-rail force monitoring model provided by an embodiment of the present application is shown. Figure 2 Another flow chart of a training method for a heavy-duty train wheel-rail force monitoring model provided by an embodiment of the present application is shown. Figure 1 and 2 As shown, it may include the following steps:

[0049] S101, obtaining actual wheel-rail forces and simulated wheel-rail forces of multiple groups of track inspection vehicles and heavy-load trains running on preset tracks, the actual wheel-rail forces including actual wheel-rail forces of the track inspection vehicles and actual wheel-rail forces of the heavy-load trains, and the simulated wheel-rail forces including simulated wheel-rail forces of the track inspection vehicles and simulated wheel-rail forces of the heavy-load trains;

[0050] S102, performing model training on each set of simulated wheel-rail forces of the track inspection vehicle and the simulated wheel-rail forces of the heavy-load train, and obtaining a mapping relationship model of the simulated wheel-rail forces of the track inspection vehicle and the heavy-load train;

[0051] S103, inputting the actual wheel-rail force of the track inspection vehicle and the actual wheel-rail force of the heavy-load train into the simulated wheel-rail force mapping relationship model to perform neural network transfer learning, and obtain a heavy-load train wheel-rail force monitoring model.

[0052] Based on the above embodiment, the actual wheel-rail forces and simulated wheel-rail forces of multiple groups of track inspection vehicles and heavy-load trains traveling on preset tracks are obtained; model training is performed on each group of simulated wheel-rail forces of the track inspection vehicles and the simulated wheel-rail forces of the heavy-load trains to obtain a simulated wheel-rail force mapping relationship model of the track inspection vehicles and the heavy-load trains; the actual wheel-rail forces of the track inspection vehicles and the actual wheel-rail forces of the heavy-load trains are input into the simulated wheel-rail force mapping relationship model to perform neural network transfer learning to obtain a heavy-load train wheel-rail force monitoring model. By training models for multiple sets of track inspection vehicle simulated wheel-rail forces and heavy-load train simulated wheel-rail forces, the mapping relationship between the heavy-load train and the track inspection vehicle can be accurately determined when the wheel-rail forces are obtained by simulation calculation. Then, the simulated wheel-rail force mapping relationship model is transferred from the simulation data to reality through transfer learning, so that the wheel-rail forces detected by the heavy-load train wheel-rail force monitoring model are more accurate, thereby improving the safety and reliability of railway transportation and reducing the probability of accidents. At the same time, the heavy-load train wheel-rail force monitoring model can also be used to monitor the wheel-rail forces generated by heavy-load trains when they are running on tracks for a long time, without the need to lay a large number of sensors for detection. This can provide a scientific basis for the maintenance and management of railway infrastructure, extend the service life of the line, and save detection costs.

[0053] In the above S101, the track inspection vehicle is a special vehicle used to detect the geometric state and unevenness of the track in order to evaluate the geometric state of the track. Wheel-rail force refers to the dynamic force of the interaction between the wheels and the track when the train is running on the track. A wheel-rail force detection sensor can be set on the track inspection vehicle, so that the actual wheel-rail force of the track inspection vehicle generated by the track inspection vehicle when it is running on the preset track can be detected by the wheel-rail force sensor.

[0054] Ground sensors can be pre-buried under the preset track. When a heavy-loaded train or a track inspection vehicle is running on the preset track, the ground sensors can detect the actual wheel-rail force of the heavy-loaded train and the actual wheel-rail force of the track inspection vehicle when the heavy-loaded train or the track inspection vehicle passes by.

[0055] Through software simulation testing, it is also possible to test and obtain the simulated wheel-rail force of the heavy-load train and the simulated wheel-rail force of the track inspection vehicle generated when the heavy-load train and the track inspection vehicle are running on the preset track.

[0056] The actual wheel-rail force and simulated wheel-rail force of each set of track inspection vehicles and heavy-load trains running on the preset track refer to the actual wheel-rail force and simulated wheel-rail force detected by the heavy-load train and the track inspection vehicle at the same position on the preset track.

[0057] In the above S102, the simulated wheel-rail force of the track inspection vehicle and the simulated wheel-rail force of the heavy-load train are obtained by simulation testing at the same position of the preset track, and the simulation standards are consistent. Therefore, each set of simulated wheel-rail force of the track inspection vehicle and the simulated wheel-rail force of the heavy-load train can be input into the model for model training, so as to obtain a simulated wheel-rail force mapping relationship model about the correspondence between the wheel-rail force generated by the track inspection vehicle and the heavy-load train when passing the same position of the preset track.

[0058] Specifically, in an example, the above S102 may include:

[0059] S1021, using the simulated wheel-rail force of the track inspection vehicle as the model input and the simulated wheel-rail force of the heavy-load train as the training label to train the deep residual network model, and obtain the simulated wheel-rail force mapping relationship model of the track inspection vehicle and the heavy-load train.

[0060] The deep residual network model is a deep learning network architecture. The basic structure is composed of multiple residual blocks stacked together. In traditional convolutional neural networks, the features learned by each convolutional layer will be passed to the next layer in turn. In the deep residual network model, each residual block can directly skip several layers of features. The advantage of this is that the original input information is retained to prevent information loss. At the same time, if there is a gap between the output of the current layer and the output of the previous layer, the residual block can adjust this gap by learning parameters, making it easier for the network to fit the residual part. Each residual block contains two convolutional layers and cross connections. Among them, the cross connection can directly skip several layers of input, thereby building a very deep network, solving the gradient explosion and gradient attenuation problems caused by the increase in the number of layers of the deep learning model. The number of network layers can be increased by continuously increasing the number of residual blocks. The residual block directly adds the input signal to the output signal, thereby introducing cross-layer connections, allowing information to flow more smoothly in the network. In addition, the residual block can also reduce the risk of overfitting and improve the generalization ability of the network.

[0061] Therefore, by training the deep residual network model, the complex mapping relationship between the wheel-rail force of the track inspection vehicle and the heavy-load train can be learned. This mapping relationship is based on a large number of simulated wheel-rail forces and can more accurately reflect the dynamic characteristics between the track inspection vehicle and the heavy-load train. The deep residual network has powerful feature extraction and nonlinear fitting capabilities, can process complex input data, and output a high-precision simulated wheel-rail force mapping relationship model, thereby improving the accuracy of the simulated wheel-rail force mapping relationship model.

[0062] In the above S103, since the simulated wheel-rail force is obtained by simulating the software, and the software cannot truly reflect its high-frequency signal when processing complex motion processes (such as: small curvature radius curves), it is impossible to accurately obtain the wheel-rail force generated by the heavy-load train or track inspection vehicle during the actual motion process. Therefore, it is necessary to migrate the simulated wheel-rail force mapping relationship model to the heavy-load train wheel-rail force monitoring model oriented to real data by adopting the neural network transfer learning method.

[0063] Specifically, in an example, the above S103 may further include:

[0064] S1031, using the actual wheel-rail force of the track inspection vehicle as the model input and the actual wheel-rail force of the heavy-load train as the training label to train the simulated wheel-rail force mapping relationship model, and obtain the heavy-load train wheel-rail force monitoring model.

[0065] By directly using actual wheel-rail force data for training, the model can learn the true mapping relationship between the wheel-rail force between the track inspection vehicle and the heavy-load train, thereby improving the accuracy of the detection results.

[0066] In one example, in order to obtain the simulated wheel-rail force more accurately, the present application also provides another implementation of the training method of the heavy-load train wheel-rail force monitoring model.

[0067] Before the above S101, the training method of the heavy-load train wheel-rail force monitoring model may further include:

[0068] S201, obtaining a track inspection vehicle structure model of a track inspection vehicle, a heavy-load train structure model of a heavy-load train, and road condition information of a preset track;

[0069] S202, constructing a track inspection wheel-rail force simulation model of the track inspection vehicle passing through a preset track and a heavy-load train wheel-rail force simulation model of the heavy-load train passing through a preset track respectively according to the track inspection vehicle structure model, the heavy-load train structure model and the road condition information;

[0070] S203, respectively determining the track inspection vehicle simulation wheel-rail force of the track inspection vehicle and the heavy-load train simulation wheel-rail force of the heavy-load train through the track inspection wheel-rail force simulation model and the heavy-load train wheel-rail force simulation model.

[0071] Based on the above implementation, by constructing accurate track inspection vehicle and heavy-load train structural models and considering the road condition information of the preset track, a more accurate track inspection wheel-rail force simulation model and a heavy-load train wheel-rail force simulation model can be constructed. Therefore, accurate track inspection vehicle simulated wheel-rail force and heavy-load train simulated wheel-rail force can be obtained based on the track inspection wheel-rail force simulation model and the heavy-load train wheel-rail force simulation model.

[0072] In the above S201, the structural model of the track inspection vehicle can be constructed based on the size data and material structure of the track inspection vehicle itself. Similarly, the structural model of the heavy-load train can also be constructed.

[0073] The road condition information may be acquired from the preset track by actual measurement. Specifically, the road condition information may include: turning radius, turning arc, track slope, etc.

[0074] In one embodiment, obtaining the road condition information of the preset track may further include:

[0075] S2011, obtaining a comprehensive line map of a preset track;

[0076] S2012, analyzing the comprehensive line map to obtain road condition information of the preset track.

[0077] In the above S2011, the comprehensive line map is a drawing of a railway network, which can reflect the geometric characteristics or linear elements of the railway, such as basic structural data such as curves, ramps, vertical curves, transition curves and superelevations. The comprehensive line map of the preset track can be obtained by consulting data.

[0078] In the above S2012, the route comprehensive map is read and the traffic information can be parsed from the route comprehensive map.

[0079] In the above S202, the road condition information and the track inspection vehicle structure model are input into the simulation software, and a track inspection wheel-rail force simulation model can be constructed through the simulation software. The model can simulate the wheel-rail force generated when the track inspection vehicle travels at various positions on the preset track.

[0080] Similarly, a heavy-load train wheel-rail force simulation model can also be constructed, which can simulate the wheel-rail force generated when the heavy-load train runs at various positions on the preset track.

[0081] In the above S203, by running the track inspection wheel-rail force simulation model, the track inspection vehicle simulated wheel-rail force at any position on the preset track can be output. Similarly, running the heavy-load train wheel-rail force simulation model can also obtain the heavy-load train simulated wheel-rail force at any position on the preset track.

[0082] In one example, in order to more accurately obtain the actual wheel-rail force of the track inspection vehicle, the present application also provides another implementation of the training method of the heavy-load train wheel-rail force monitoring model.

[0083] Before the above S101, the training method of the heavy-load train wheel-rail force monitoring model may further include:

[0084] S301, obtaining wheel-rail forces measured by the track inspection vehicle and the wheel-rail forces measured by sensors when the track inspection vehicle passes through a preset track;

[0085] S302, using the wheel-rail force measured by the sensor as a standard value, calibrating the wheel-rail force measured by the track inspection vehicle to obtain the actual wheel-rail force of the track inspection vehicle.

[0086] Based on the above embodiment, the wheel-rail force measured by the track inspection vehicle and the wheel-rail force measured by the sensor are obtained when the track inspection vehicle passes through the preset track, and then the wheel-rail force measured by the sensor is used as the standard value to correct the wheel-rail force measured by the track inspection vehicle. This can reduce the error caused by vibration and weather in the force measuring wheels of the track inspection vehicle during long-term operation, and improve the accuracy of the actual wheel-rail force of the track inspection vehicle.

[0087] In the above S301, the wheel-rail force detected by the vehicle sensor of the track inspection vehicle when the track inspection vehicle passes through the preset track is the wheel-rail force measured by the track inspection vehicle, which can be directly obtained from the sensor on the track inspection vehicle.

[0088] The wheel-rail force detected by the track can be obtained directly from the ground sensors at the preset track, and the sensor measures the wheel-rail force when the track inspection vehicle passes through the preset track.

[0089] Specifically, the above S301 may also include:

[0090] S3011, obtaining the wheel-rail force of the track inspection vehicle when the track inspection vehicle passes through the preset track by using the force measuring wheel pair technology;

[0091] S3012, obtaining the sensor wheel-rail force of the track inspection vehicle passing through the preset track through the ground sensor, and the ground sensor is connected to the preset track.

[0092] In the above S3011, the force measuring wheelset technology is a technology used to measure the force between wheel and rail, which can directly measure the force between wheel and rail with high measurement accuracy. Therefore, the force measuring wheelset technology can be used to directly detect the wheel-rail force measured by the track inspection vehicle passing through the preset track.

[0093] In the above S3012, the ground sensor is connected to the preset track. When the track inspection vehicle travels on the preset track, the ground sensor can directly obtain the sensor wheel-rail force generated between the track inspection vehicle and the preset track from the preset track.

[0094] In the above S302, since the dynamometric wheelset of the track inspection vehicle is susceptible to vibration and weather during long-term operation, errors are easily generated. Therefore, it is necessary to first calibrate the dynamometric wheelset of the track inspection vehicle using a ground sensor. The wheel-rail force measured by the ground sensor is used as the standard value to calibrate the wheel-rail force of the track inspection vehicle, thereby eliminating the errors generated by the dynamometric wheelset of the track inspection vehicle under long-term use.

[0095] Specifically, the above S302 may also include:

[0096] S3021, input the wheel-rail force measured by the track inspection vehicle and the wheel-rail force measured by the sensor into the BP neural network model for calculation to obtain the actual wheel-rail force.

[0097] In the above S3021, the BP neural network has good robustness and fault tolerance, which means that even if there are certain errors or noises in the input data, the BP neural network can output relatively accurate results. The wheel-rail force measured by the track inspection vehicle is used as input data, and the wheel-rail force measured by the sensor is used as a label. The actual wheel-rail force of the track inspection vehicle is calculated through the BP neural network model.

[0098] Embodiment 2;

[0099] Based on the same inventive concept, this embodiment further provides a training device 200 for a heavy-duty train wheel-rail force monitoring model on the basis of Embodiment 1, as shown in FIG. Figure 3 As shown, it includes:

[0100] The first acquisition module 210 is used to obtain the actual wheel-rail force and simulated wheel-rail force of multiple groups of track inspection vehicles and heavy-load trains running on the preset track, the actual wheel-rail force includes: the actual wheel-rail force of the track inspection vehicle and the actual wheel-rail force of the heavy-load train, and the simulated wheel-rail force includes: the simulated wheel-rail force of the track inspection vehicle and the simulated wheel-rail force of the heavy-load train;

[0101] The training module 220 is used to perform model training on each set of track inspection vehicle simulated wheel-rail force and heavy-load train simulated wheel-rail force to obtain a mapping relationship model of the simulated wheel-rail force of the track inspection vehicle and the heavy-load train;

[0102] The transfer learning module 230 is used to input the actual wheel-rail force of the track inspection vehicle and the actual wheel-rail force of the heavy-load train into the simulated wheel-rail force mapping relationship model to perform neural network transfer learning and obtain a heavy-load train wheel-rail force monitoring model.

[0103] Based on the above embodiment, the actual wheel-rail forces and simulated wheel-rail forces of multiple groups of track inspection vehicles and heavy-load trains traveling on preset tracks are obtained through the first acquisition module 210; the training module 220 performs model training on each group of simulated wheel-rail forces of the track inspection vehicle and the simulated wheel-rail forces of the heavy-load train to obtain a simulated wheel-rail force mapping relationship model of the track inspection vehicle and the heavy-load train; the transfer learning module 230 inputs the actual wheel-rail forces of the track inspection vehicle and the actual wheel-rail forces of the heavy-load train into the simulated wheel-rail force mapping relationship model to perform neural network transfer learning to obtain a heavy-load train wheel-rail force monitoring model. Based on this, by training models of multiple sets of track inspection vehicle simulated wheel-rail forces and heavy-load train simulated wheel-rail forces, the mapping relationship between the heavy-load train and the track inspection vehicle can be accurately determined when the wheel-rail forces are obtained by simulation calculation. Then, through transfer learning, the simulated wheel-rail force mapping relationship model is transferred from the simulation data to reality, so that the wheel-rail forces detected by the heavy-load train wheel-rail force monitoring model are more accurate, the safety and reliability of railway transportation are improved, and the probability of accidents is reduced. At the same time, the heavy-load train wheel-rail force monitoring model can also be used to monitor the wheel-rail forces generated by heavy-load trains when they are running on tracks for a long time, without the need to lay a large number of sensors for detection, which can provide a scientific basis for the maintenance and management of railway infrastructure, extend the service life of the line, and save detection costs.

[0104] Optionally, the training device 200 for the heavy-load train wheel-rail force monitoring model may further include:

[0105] The second acquisition module is used to acquire the track inspection vehicle structure model of the track inspection vehicle, the heavy-load train structure model of the heavy-load train, and the road condition information of the preset track;

[0106] A model building module is used to build a track inspection vehicle wheel-rail force simulation model when the track inspection vehicle passes through a preset track, and a heavy-load train wheel-rail force simulation model when the heavy-load train passes through a preset track, respectively, according to the track inspection vehicle structure model, the heavy-load train structure model and road condition information;

[0107] The determination module is used to determine the track inspection vehicle simulation wheel-rail force and the heavy-load train simulation wheel-rail force of the track inspection vehicle and the heavy-load train simulation wheel-rail force respectively through the track inspection wheel-rail force simulation model and the heavy-load train wheel-rail force simulation model.

[0108] Optionally, the training device 200 for the heavy-load train wheel-rail force monitoring model may further include:

[0109] The third acquisition module is used to obtain a comprehensive line map of the preset track;

[0110] The parsing module is used to parse the comprehensive line map and obtain the road condition information of the preset track.

[0111] Optionally, the training device 200 for the heavy-load train wheel-rail force monitoring model may further include:

[0112] A fourth acquisition module is used to obtain the wheel-rail force measured by the track inspection vehicle and the wheel-rail force measured by the sensor when the track inspection vehicle passes through the preset track;

[0113] The correction module is used to correct the wheel-rail force measured by the track inspection vehicle using the wheel-rail force measured by the sensor as the standard value to obtain the actual wheel-rail force.

[0114] Optionally, the correction module may include:

[0115] The calculation unit is used to input the wheel-rail force measured by the track inspection vehicle and the wheel-rail force measured by the sensor into the BP neural network model for calculation to obtain the actual wheel-rail force.

[0116] Optionally, the fourth acquisition module may include:

[0117] The first acquisition unit is used to acquire the wheel-rail force of the track inspection vehicle when the track inspection vehicle passes through the preset track by using the force measuring wheel pair technology;

[0118] The second acquisition unit is used to acquire the sensor wheel-rail force of the track inspection vehicle passing through the preset track through the ground sensor, and the ground sensor is connected to the preset track.

[0119] Optionally, the training module 220 may include:

[0120] The first training unit is used to train the deep residual network model using the simulated wheel-rail force of the track inspection vehicle as the model input and the simulated wheel-rail force of the heavy-load train as the training label, so as to obtain the simulated wheel-rail force mapping relationship model of the track inspection vehicle and the heavy-load train.

[0121] Optionally, the transfer learning module 230 may include:

[0122] The second training unit is used to use the actual wheel-rail force of the track inspection vehicle as the model input and the actual wheel-rail force of the heavy-load train as the training label to train the simulated wheel-rail force mapping relationship model to obtain the heavy-load train wheel-rail force monitoring model.

[0123] In Example 2, each process of the above-mentioned Example 1 can be implemented and the same technical effect can be achieved. To avoid repetition, it will not be described here.

[0124] Embodiment 3:

[0125] Figure 4 A schematic diagram of the hardware structure of a training device for a heavy-load train wheel-rail force monitoring model provided by an embodiment of the present invention is shown.

[0126] The training device for the heavy-load train wheel-rail force monitoring model may include a processor 301 and a memory 302 storing computer program instructions.

[0127] Specifically, the processor 301 may include a central processing unit (CPU), or an application specific integrated circuit (ASIC), or may be configured to implement one or more integrated circuits of the embodiment of the present invention.

[0128] The memory 302 may include a large capacity memory for data or instructions. By way of example and not limitation, the memory 302 may include a hard disk drive (HDD), a floppy disk drive, a flash memory, an optical disk, a magneto-optical disk, a magnetic tape, or a universal serial bus (USB) drive or a combination of two or more of these. In appropriate cases, the memory 302 may include a removable or non-removable (or fixed) medium. In appropriate cases, the memory 302 may be inside or outside the integrated gateway disaster recovery device. In a specific embodiment, the memory 302 is a non-volatile solid-state memory.

[0129] In certain embodiments, the memory 302 may include a read-only memory (ROM), a random access memory (RAM), a magnetic disk storage media device, an optical storage media device, a flash memory device, an electrical, optical or other physical / tangible memory storage device. Thus, typically, the memory 302 includes one or more tangible (non-transitory) computer-readable storage media (e.g., a memory device) encoded with software including computer-executable instructions, and when the software is executed (e.g., by one or more processors 301), it is operable to perform the operations described with reference to the method according to an aspect of the present application.

[0130] The processor 301 implements any one of the training methods for the heavy-load train wheel-rail force monitoring model in the above embodiments by reading and executing the computer program instructions stored in the memory 302 .

[0131] In one example, the training device for the heavy-load train wheel-rail force monitoring model may further include a communication interface 303 and a bus 304. As shown in the figure, the processor 301, the memory 302, and the communication interface 303 are connected through the bus 304 and communicate with each other.

[0132] The communication interface 303 is mainly used to implement the communication between the modules, devices, units and / or equipment in the embodiment of the present invention.

[0133] The bus 304 includes hardware, software or both. For example and not limitation, the bus 304 may include an accelerated graphics port (AGP) or other graphics bus, an enhanced industry standard architecture (EISA) bus, a front-side bus (FSB), a hypertransport (HT) interconnect, an industry standard architecture (ISA) bus, a wireless bandwidth interconnect, a low pin count (LPC) bus, a memory bus, a microchannel architecture (MCA) bus, a peripheral control interconnect (PCI) bus, a PCI-Express (PCI-X) bus, a serial advanced technology attachment (SATA) bus, a video electronics standard association local (VLB) bus or other suitable bus or a combination of two or more of these. Where appropriate, the bus 304 may include one or more buses 304. Although the present application embodiment describes and shows a specific bus 304, the present application considers any suitable bus 304 or interconnection.

[0134] The training device of the heavy-load train wheel-rail force monitoring model can be based on the current training method of the heavy-load train wheel-rail force monitoring model, thereby realizing the combination of Figure 1 , 3 The invention describes a training method for a heavy-load train wheel-rail force monitoring model and a training device 200 for a heavy-load train wheel-rail force monitoring model.

[0135] In addition, the embodiment of the present application also provides a computer program product, including computer program instructions, and when the computer program product is executed by the processor 301, the steps and corresponding contents of the aforementioned method embodiment can be implemented.

[0136] It should be clear that the present application is not limited to the specific configuration and processing described above and shown in the figures. For the sake of simplicity, a detailed description of the known method is omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of the present application is not limited to the specific steps described and shown, and those skilled in the art can make various changes, modifications and additions, or change the order between the steps after understanding the spirit of the present application.

[0137] The functional blocks shown in the above-described block diagram can be implemented as hardware, software, firmware or a combination thereof. When implemented in hardware, it can be, for example, an electronic circuit, an application specific integrated circuit (ASIC), appropriate firmware, a plug-in, a function card, etc. When implemented in software, the elements of the present application are programs or code segments that are used to perform the required tasks. The program or code segment can be stored in a machine-readable medium, or transmitted on a transmission medium or a communication link by a data signal carried in a carrier wave. "Machine-readable medium" can include any medium capable of storing or transmitting information. Examples of machine-readable media include electronic circuits, semiconductor memory devices, ROM, flash memory, erasable ROM (EROM), floppy disks, CD-ROMs, optical disks, hard disks, optical fiber media, radio frequency (RF) links, etc. The code segment can be downloaded via a computer network such as the Internet, an intranet, etc.

[0138] It should also be noted that the exemplary embodiments mentioned in this application describe some methods or systems based on a series of steps or devices. However, this application is not limited to the order of the above steps, that is, the steps can be performed in the order mentioned in the embodiment, or in a different order from the embodiment, or several steps can be performed simultaneously.

[0139] Aspects of the present disclosure are described above with reference to the flowcharts and / or block diagrams of the methods, devices (systems) and computer program products according to the embodiments of the present disclosure. It should be understood that each box in the flowchart and / or block diagram and the combination of boxes in the flowchart and / or block diagram can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable heavy-duty train wheel-rail force monitoring model training method, device and equipment to produce a machine so that these instructions executed by the processor of the computer or other programmable heavy-duty train wheel-rail force monitoring model training method, device and equipment enable the implementation of the functions / actions specified in one or more boxes of the flowchart and / or block diagram. Such a processor can be, but is not limited to, a general-purpose processor, a special-purpose processor, a special application processor or a field programmable logic circuit. It can also be understood that each box in the block diagram and / or flowchart and the combination of boxes in the block diagram and / or flowchart can also be implemented by dedicated hardware that performs the specified function or action, or can be implemented by a combination of dedicated hardware and computer instructions.

[0140] The above is only a specific implementation of the present application. Those skilled in the art can clearly understand that for the convenience and simplicity of description, the specific working processes of the systems, modules and units described above can refer to the corresponding processes in the aforementioned method embodiments, and will not be repeated here. It should be understood that the protection scope of the present application is not limited to this. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed in this application, and these modifications or replacements should be included in the protection scope of this application.

Claims

1. A training method for a heavy-load train wheel-rail force monitoring model, characterized in that: The method comprises: Acquire actual wheel-rail forces and simulated wheel-rail forces of multiple groups of track inspection vehicles and heavy-load trains running on preset tracks, wherein the actual wheel-rail forces include: actual wheel-rail forces of the track inspection vehicles and actual wheel-rail forces of the heavy-load trains, and the simulated wheel-rail forces include: simulated wheel-rail forces of the track inspection vehicles and simulated wheel-rail forces of the heavy-load trains; Performing model training on each group of the track inspection vehicle simulated wheel-rail force and the heavy-load train simulated wheel-rail force to obtain a mapping relationship model of the simulated wheel-rail force of the track inspection vehicle and the heavy-load train; The actual wheel-rail force of the track inspection vehicle and the actual wheel-rail force of the heavy-load train are input into the simulated wheel-rail force mapping relationship model to perform neural network transfer learning, so as to obtain a heavy-load train wheel-rail force monitoring model.

2. The method according to claim 1, characterized in that Before obtaining the actual wheel-rail forces and simulated wheel-rail forces of multiple groups of track inspection vehicles and heavy-load trains traveling on preset tracks, the method further includes: Acquire a track inspection vehicle structure model of the track inspection vehicle, a heavy-load train structure model of the heavy-load train, and road condition information of the preset track; According to the track inspection vehicle structure model, the heavy-load train structure model and the road condition information, a track inspection wheel-rail force simulation model of the track inspection vehicle passing through the preset track and a heavy-load train wheel-rail force simulation model of the heavy-load train passing through the preset track are respectively constructed; The track inspection vehicle simulated wheel-rail force of the track inspection vehicle and the heavy-load train simulated wheel-rail force of the heavy-load train are determined respectively by the track inspection vehicle wheel-rail force simulation model and the heavy-load train wheel-rail force simulation model.

3. The method according to claim 2, characterized in that Before acquiring the track inspection vehicle structure model of the track inspection vehicle, the heavy-load train structure model of the heavy-load train, and the road condition information of the preset track, the method further includes: Obtaining a comprehensive line map of the preset track; The route comprehensive map is analyzed to obtain the road condition information of the preset track.

4. The method according to claim 1, characterized in that Before obtaining the actual wheel-rail forces and simulated wheel-rail forces of multiple groups of track inspection vehicles and heavy-load trains traveling on preset tracks, the method further includes: Obtaining the wheel-rail force measured by the track inspection vehicle and the wheel-rail force measured by the sensor when the track inspection vehicle passes through the preset track; The wheel-rail force measured by the sensor is used as a standard value to calibrate the wheel-rail force measured by the track inspection vehicle to obtain the actual wheel-rail force of the track inspection vehicle.

5. The method according to claim 4, characterized in that The step of obtaining the wheel-rail force measured by the track inspection vehicle and the wheel-rail force measured by the sensor when the track inspection vehicle passes through the preset track includes: The wheel-track force of the track inspection vehicle when the track inspection vehicle passes through the preset track is obtained by using the force measuring wheel pair technology; The sensor wheel-rail force of the track inspection vehicle passing through the preset track is obtained through a ground sensor, and the ground sensor is connected to the preset track.

6. The method according to claim 4, characterized in that The wheel-rail force measured by the sensor is used as a standard value to calibrate the wheel-rail force measured by the track inspection vehicle to obtain the actual wheel-rail force of the track inspection vehicle, including: The wheel-rail force measured by the track inspection vehicle and the wheel-rail force measured by the sensor are input into the BP neural network model for calculation to obtain the actual wheel-rail force.

7. The method according to claim 1, characterized in that The model training is performed on each group of the track inspection vehicle simulated wheel-rail force and the heavy-load train simulated wheel-rail force to obtain the simulated wheel-rail force mapping relationship model of the track inspection vehicle and the heavy-load train, including: The simulated wheel-rail force of the track inspection vehicle is used as the model input, and the simulated wheel-rail force of the heavy-load train is used as the training label to train the deep residual network model, so as to obtain the simulated wheel-rail force mapping relationship model of the track inspection vehicle and the heavy-load train.

8. The method according to claim 1, characterized in that The step of inputting the actual wheel-rail force of the track inspection vehicle and the actual wheel-rail force of the heavy-load train into the simulated wheel-rail force mapping relationship model to perform neural network transfer learning to obtain a heavy-load train wheel-rail force monitoring model comprises: The actual wheel-rail force of the track inspection vehicle is used as a model input, and the actual wheel-rail force of the heavy-load train is used as a training label to perform model training on the simulated wheel-rail force mapping relationship model to obtain the heavy-load train wheel-rail force monitoring model.

9. A training device for a heavy-load train wheel-rail force monitoring model, characterized in that: include: The first acquisition module is used to obtain the actual wheel-rail force and simulated wheel-rail force of multiple groups of track inspection vehicles and heavy-load trains running on preset tracks, wherein the actual wheel-rail force includes: the actual wheel-rail force of the track inspection vehicle and the actual wheel-rail force of the heavy-load train, and the simulated wheel-rail force includes: the simulated wheel-rail force of the track inspection vehicle and the simulated wheel-rail force of the heavy-load train; A training module, used for performing model training on each group of simulated wheel-rail forces of the track inspection vehicle and the simulated wheel-rail forces of the heavy-load train, to obtain a mapping relationship model of the simulated wheel-rail forces of the track inspection vehicle and the heavy-load train; The transfer learning module is used to input the actual wheel-rail force of the track inspection vehicle and the actual wheel-rail force of the heavy-load train into the simulated wheel-rail force mapping relationship model to perform neural network transfer learning to obtain a heavy-load train wheel-rail force monitoring model.

10. A training device for a heavy-load train wheel-rail force monitoring model, characterized in that: The device comprises: a processor and a memory storing computer program instructions; When the processor executes the computer program instructions, the training method for the heavy-load train wheel-rail force monitoring model as described in any one of claims 1-8 is implemented.

Images