Method and System for Model Data Processing of Train Energy Absorption Structure in Collision Simulation

Abstract

The present invention relates to the technical field of train crashworthiness performance analysis, and discloses a method and system for processing model data of a train energy absorption structure in collision simulation to achieve cost reduction and efficiency improvement and ensure reliability. The method includes: using the force-displacement curve of each sample in the first dataset as the input and output of a self-supervised network model, and extracting the intermediate layer hidden variables respectively corresponding to each force-displacement curve after dimensionality reduction in the self-supervised network model; then training, validating and testing a new network model with the structural parameters of the energy absorption structure as the input and the intermediate layer hidden variables as the output according to the second dataset; combining the new network model with the decoder part of the self-supervised network model to obtain a target model for predicting the force-displacement curve based on the structural parameters; embedding the target model into the coupled train collision dynamics model to predict and output the force-displacement curve required for collision simulation according to the input data of the dynamic simulation environment.

Description

Technical Field

[0001] The present invention relates to the technical field of train crashworthiness analysis, and particularly to a method and system for processing model data of a train energy absorption structure in a collision simulation. Background Art

[0002] The passive protection of trains involves complex mechanical properties of geometric nonlinear material contact, deformation, and coupling. It is generally divided into five stages: coupler buffer device, crush tube, coupler shear bolt, end energy absorption structure, and carbody energy absorption. Once the carbody energy absorption part occurs, it will cause the deformation of the train and the casualties of passengers. Therefore, the analysis and design of train crashworthiness are crucial. Especially the crashworthiness design method of formation trains.

[0003] In the design of modern urban rail trains, impact energy management is a key link to ensure the passive safety of trains. The multi-level energy management strategy aims to maximize the protection of the passenger area, ensure its structural integrity, and reduce the harm to passengers. In addition, in addition to these energy management measures, it is also essential to conduct a detailed formation crashworthiness analysis to ensure that the entire train formation has sufficient crashworthiness and safety in a collision event. Through these comprehensive measures, the design of modern urban rail trains can better cope with potential collision risks and improve the overall safety performance.

[0004] The traditional finite element method has high accuracy in calculating the crashworthiness of formation trains, but its calculation efficiency is low. If the finite element simulation is simply embedded in the energy absorption structure and the crashworthiness analysis of formation trains, it will be difficult to achieve the rapid mapping and iterative optimization of structural parameters and the crashworthiness of formation trains. Some existing studies have introduced dynamic methods for the crashworthiness analysis of energy absorption structures and formation trains, but they are prone to deviation in the process of parameter inverse solution, and the efficiency improvement is limited. Summary of the Invention

[0005] The purpose of the present invention is to disclose a method and system for processing model data of a train energy absorption structure in a collision simulation to achieve cost reduction, efficiency improvement, and ensure reliability.

[0006] To achieve the above object, the method for processing model data of a train energy absorption structure in a collision simulation disclosed by the present invention includes:

[0007] Step S1, obtaining force-displacement curves of different energy absorption structures based on finite element simulation to construct a first data set;

[0008] Step S2, using the force-displacement curve of each sample in the first data set as the input and output of a self-supervised network model, and extracting the intermediate layer hidden variables respectively corresponding to each force-displacement curve after dimensionality reduction in the self-supervised network model;

[0009] Step S3: Construct a second data set based on the intermediate layer latent variables respectively corresponding to the structural parameters of different energy absorption structures in the self-supervised network model;

[0010] Step S4: Train, validate, and test a new network model with the structural parameters of the energy absorption structure as the input and the intermediate layer latent variable as the output according to the second data set;

[0011] Step S5: Combine the new network model with the decoder part of the self-supervised network model to obtain a target model for predicting the force-displacement curve based on the structural parameters, and the weights, offset value information, and processing logic of the combined decoder part are consistent with the decoder weights, offset value information, and processing logic obtained in Step S2;

[0012] Step S6: Embed the target model into the coupled train collision dynamics model to predict and output the force-displacement curve required for collision simulation according to the input data of the dynamics simulation environment.

[0013] Preferably, in Step S1, the finite element simulation specifically is to construct and experimentally verify a finite element parametric model of the train energy absorption structure in a finite element simulation software. The finite element parametric model encapsulates the entire process code of geometric parameter driving, material association, load application, mesh generation, and calculation submission, realizing the automatic mapping of "parameter input - simulation output", and the first data set is constructed based on the finite element parametric model by the DOE (Design Of Experiments) method.

[0014] Preferably, in Step S2, the force-displacement data of each sample is converted into a 2×1000-dimensional matrix, and then four algorithms of AE, RNN-AE, LSTM-AE, and GRU-AE network models are respectively used for model training. The prediction accuracies of different models are compared and analyzed, and the network model with the highest accuracy of the predicted output curve is used as the self-supervised network model.

[0015] Preferably, in Step S4, the new network model uses a CNN network model. The energy absorption structure specifically is the anti-climbing device in the coupler draft gear. The structural parameters include: the parting surface diameter of the friction tube and the friction shoe, the pre-tightening pressure of the friction shoe, the dynamic friction coefficient of the friction rubber, the pre-connection length of the friction tube and the friction shoe, the thickness coefficient of the shoe, and the taper coefficient.

[0016] Preferably, in the coupled train collision dynamics model, the coupler draft gear includes a coupler and the anti-climbing device. The constitutive relationship of the local plastic deformation of the structure of the coupler draft gear is defined by a generalized spring, and the mechanical characteristic curves of the energy absorption structure and the coupler are constructed using a nonlinear hysteresis mathematical model. The nonlinear hysteresis mathematical model is described in the following three stages:

[0017] The calculation equation in the loading stage is as follows: ; where is the impedance force when the coupler or anti-climbing device is loaded; is the initial pressure of the coupler or anti-climbing device; is the sliding friction force; is the static friction force; represents the relative longitudinal speed between the moving end and the stationary end of the coupler buffer; represents the Stribeck critical judgment speed;

[0018] The initial pressure of the coupler and anti-climbing device will affect the friction coefficient at the beginning of the impact. In the actual calculation process, it is often calculated as a constant and rewritten in the form of stiffness as: ; where is the linear stiffness of the coupler buffer loading; is the stiffness under the simultaneous action of sliding and static friction; represents the displacement of the buffer along the longitudinal direction of the train during the loading stage;

[0019] In the unloading stage, the impedance force of the coupler buffer and the stroke can be regarded as linearly related, and the calculation method is: ; is the linear stiffness in the unloading stage;

[0020] Among them, there are fluctuations between loading and unloading. The moving section and the stationary section of the coupler buffer are relatively stationary, and the relative speed between them is 0, thus causing the zero-speed stage;

[0021] When the stationary end and the moving end of the coupler buffer system are in the stage of relative speed being zero, the direction of the speed will change, resulting in non-uniqueness of the magnitude and direction of the instantaneous force in the numerical simulation; In order to accurately determine the friction damping characteristics at the zero-speed moment, so that the magnitude of the impedance force in the calculation process has a unique solution and avoid the fluctuation phenomenon in the simulation output, the mathematical model of the instantaneous impedance force at the zero-speed moment is established as follows:

[0022] ;

[0023] where is the impedance force of the coupler buffer device at is the impedance force of the coupler buffer device at is the coupler buffer displacement of the coupler buffer device at is the coupler buffer displacement of the coupler buffer device at is the geometric control parameter for controlling the change of the loading and unloading boundary; is the boundary force of the coupler buffer characteristic curve;

[0024] By instantaneous displacement and instantaneous displacement to determine the boundary force , when , take the impedance force in the loading stage, when , take the impedance force in the unloading stage.

[0025] Preferably, in steps S2 and S5, before model training, normalize the force-displacement curve and structural parameters respectively. During the normalization process, directly normalize all values by dividing them by a maximum value. The calculation formula is:

[0026] ; ;

[0027] In the formula: and are the normalized value and the true value of the i-th input structural parameter corresponding to the j-th DOE calculation result; max is the maximum true value of the i-th input structural parameter in all DOE calculation results; and are the normalized value and the true value of the k-th point on the impact force-displacement curve in the j-th DOE calculation result respectively; max is the maximum true value of the i-th crashworthiness index in all DOE calculation results.

[0028] Preferably, the present invention further includes: in the collision simulation, taking the structural parameters of the end friction energy absorption structure as input parameters, and taking the vertical displacement and acceleration of each car body of the multi - formation train as output, to study the influence of the structural parameters of the train end friction energy absorption structure on the pose optimization of the multi - formation train; and during the optimization process, taking the minimum vertical displacement and acceleration of the car body as the optimization objectives, and obtaining the best combination of energy absorption structure parameters through the method of optimal decision - making by the Pareto solution set, so that the vertical displacement and acceleration of the train reach the minimum under collision or other dynamic load conditions.

[0029] Optionally, the specific formula of the optimization objective is as follows:

[0030] ;

[0031] In the formula, represents the acceleration of the i-th stationary car; represents the acceleration of the j-th stationary car; represents the vertical lift of the i-th stationary car; represents the vertical lift of the j-th sports car; is the parting surface diameter of the friction tube and the friction bearing bush for friction energy absorption; is the bearing bush thickness coefficient; is the pre-tightening pressure of the friction bearing bush; is the pre-connection length between the friction tube and the friction bearing bush; is the dynamic friction coefficient of the friction rubber; is the taper coefficient.

[0032] To achieve the above object, the present invention also discloses a model data processing system of a train energy absorption structure in collision simulation, including a memory, a processor, and a computer program stored on the memory and operable on the processor. When the processor executes the computer program, the above method is implemented.

[0033] The present invention has the following beneficial effects:

[0034] 1. In step S2, predicting the force-displacement curve with the force-displacement curve, the hidden variable in the middle layer is equivalent to the cipher book for sending telegrams, shared by the encoder and the decoder, and can be regarded as the encoded value after dimensionality reduction of the high-dimensional time-series data of the force-displacement curve.

[0035] 2. In step S4, both the input and output data of the model are non-time-series data, and a commonly used model can achieve a reliable effect.

[0036] 3. In order to analyze the energy absorption characteristics of the friction energy absorption structure under quasi-static compression, a WED-600 type electro-hydraulic servo universal testing machine was used for quasi-static compression testing. During the test, the middle beam was driven at a speed of 5 mm / min to compress the hybrid friction energy absorption structure. After the test system was started, the middle beam would compress the hybrid friction energy absorption structure downward at a constant speed until the test displacement reached 120 mm, the test stopped and was automatically unloaded, and then the corresponding total energy absorption was calculated according to the load-displacement curve obtained from the test acquisition system. In order to verify the accuracy of the combined target model of the present invention, the force-displacement curves and the measured crashworthiness indexes obtained from multiple groups of combined target models based on the present invention and finite element simulations were compared; it was found that the errors of the crashworthiness indexes obtained by the method of the present invention were very small, the maximum error was only 2.65%, and the overall error was less than 5%. Thus, after the combined target model of the present invention is applied to the collision dynamics model of the formation train, it can be used to evaluate the crashworthiness of the train without the participation of finite element model software and guide the parameter optimization of the energy absorption structure.

[0037] Next, the present invention will be described in further detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] The accompanying drawings, which form a part of this application, are used to provide a further understanding of the present invention. The schematic embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:

[0039] Figure 1 is a schematic flowchart of a method for processing model data of a train energy absorption structure disclosed in Embodiment 1 of the present invention in a collision simulation.

[0040] Figure 2 is a schematic diagram of the hysteresis behavior of the continuous loading and unloading process of a formation train energy absorption structure in a collision simulation disclosed in Embodiment 1 of the present invention. Detailed Embodiments

[0041] The embodiments of the present invention will be described in detail below with reference to the accompanying drawings. However, the present invention can be implemented in many different ways defined and covered by the claims.

[0042] Embodiment 1

[0043] This embodiment discloses a method for processing model data of a train energy absorption structure in a collision simulation.

[0044] As Figure 1 shown, the process of the method in this embodiment includes the following steps:

[0045] Step S1: Obtain the force-displacement curves of different energy absorption structures based on finite element simulation to construct a first data set.

[0046] In this step, the finite element simulation is specifically: in a finite element simulation software, a finite element parametric model of the train energy absorption structure is constructed and experimentally verified. The finite element parametric model encapsulates the entire process code of geometric parameter driving, material association, load application, mesh generation, and calculation submission, realizing the automatic mapping of "parameter input - simulation output". And the first data set is constructed based on the finite element parametric model by the DOE method.

[0047] Preferably, the Latin hypercube sampling method can be used for sampling in this embodiment. This method can evenly distribute sample points in a multi-dimensional space, effectively covering the entire parameter space, reducing sample deviation, and improving the efficiency and accuracy of the optimization process. To ensure the accuracy of sample calculation, the sampling number of DOE can be set to 1000 groups. This sample size can fully capture the influence of the change of input parameters on the system response and improve the accuracy of model prediction.

[0048] In this step, the so-called different energy absorption structures refer to many structures with similar structures but with specific parameter values that are not exactly the same in order to seek the optimal parameter combination. Optionally, the energy absorption structure can be specifically the anti-climbing device in the coupler draft gear, and the corresponding structure parameters include: the parting surface diameter of the friction tube and the friction shoe, the pre-tightening pressure of the friction shoe, the dynamic friction coefficient of the friction rubber, the pre-connection length of the friction tube and the friction shoe, the thickness coefficient of the shoe and the taper coefficient.

[0049] Step S2: Use the force-displacement curve of each sample in the first dataset as the input and output of the self-supervised network model, and extract the intermediate layer hidden variables corresponding to each force-displacement curve after dimensionality reduction in the self-supervised network model.

[0050] An autoencoder is an unsupervised learning neural network model used to learn the low-dimensional representation of data. It consists of an encoder and a decoder. By compressing the input data into a lower-dimensional encoded representation and then reconstructing it back to the original data through the decoder. For this reason, the self-supervised network model in this step can be implemented using an autoencoder. Before training the self-supervised network model of the force-displacement curve, perform normalization processing on the force-displacement curve and the structure parameter data.

[0051] Since the force-displacement curve is a high-dimensional time-series curve, the biggest difference between it and the structure parameter data is that it is arranged horizontally along the last dimension. This dimension represents different structure parameter types in the structure parameters, but in the force-displacement curve, it represents the time-series data of a specific DOE calculation result. If we follow the MinMaxScaler algorithm of the sklearn library used for structure parameters to perform data normalization, it will cause the force-displacement curve data to be normalized to the maximum and minimum values in the way of column index (the second-to-last dimension), which will lose the time-series information of the force-displacement curve, resulting in the possibility that the maximum value 1 or the minimum value 0 may appear simultaneously after normalizing a curve, thus turning the originally continuous curve into discrete information. Therefore, considering that the impact force-displacement curves are stacked along the rows and they all belong to the same class, directly divide all values by a maximum value during the normalization process, so that all DOE impact force-displacement curves can well retain the spatial time-series continuity. The normalization method of the impact force-displacement curve is as follows:

[0052] ; ;

[0053] In the formula: and are the normalized value and the true value of the i-th input structure parameter corresponding to the j-th group of DOE calculation results; max is the maximum true value of the i-th input structure parameter among all DOE calculation results; and are the normalized value and the true value of the k-th point on the impact force-displacement curve in the j-th group of DOE calculation results, respectively; max is the maximum true value of the i-th crashworthiness index among all DOE calculation results.

[0054] In order to obtain the intermediate layer hidden variables of the force-displacement curve in the self-supervised network model, the force-displacement data of each sample is a 2*1000-dimensional matrix, and then four algorithms of AE, RNN-AE, LSTM-AE and GRU-AE network models are used for model training. The input and output of the model are the force-displacement curve itself, and then the prediction accuracies of different models are compared and analyzed. The network model with the highest accuracy of the predicted output curve is used as the self-supervised network model. During the training process, optionally, the dimension of the intermediate layer hidden variable can be set to 6; the first data set can be divided into a training set, a test set and a validation set according to the ratio of 7:2:1. Further, in order to improve the model performance, the methods of early stopping of the learning rate and dynamic adjustment of the learning rate are also used to optimize the learning rate and the training depth, and the Adam optimization algorithm can be used to train and optimize the model.

[0055] In this step, when the self-supervised network model of the force-displacement curve and the deep learning model of the structural parameters and the intermediate layer hidden variables are successfully trained, the weights of the decoder part of the self-supervised network model of the force-displacement curve are frozen. Among them, the intermediate layer hidden variable is the input of the decoder part, and the decoder is used to decode the intermediate layer hidden variable into a force-displacement curve; on the contrary, the encoder is used to reduce the dimension of the high-dimensional time series data of the force-displacement curve into the intermediate layer hidden variable.

[0056] Step S3: Construct a second data set according to the intermediate layer hidden variables respectively corresponding to the structural parameters of different energy absorption structures in the self-supervised network model.

[0057] Step S4: Train, validate and test a new network model with the structural parameters of the energy absorption structure as the input and the intermediate layer hidden variable as the output according to the second data set.

[0058] In this step, since neither the structural parameters nor the intermediate layer hidden variables are time series data, the CNN network can be used to train the interaction mechanism and the correlation mechanism between the structural parameters and the intermediate layer hidden variables.

[0059] Step S5: Combine the new network model with the decoder part of the self-supervised network model to obtain a target model for predicting the force-displacement curve based on the structural parameters. The weights, offset value information, and processing logic of the combined decoder part are consistent with the decoder weights, offset value information, and processing logic obtained in Step S2.

[0060] In this step, the weight is a parameter connecting neurons and is used to adjust the influence of input data; the offset value is the bias parameter of each neuron and is used to adjust the activation threshold of the neuron. Both are industry terms and will not be elaborated here.

[0061] Step S6: Embed the target model into the coupled train collision dynamics model to predict and output the force-displacement curve required for collision simulation based on the input data of the dynamics simulation environment.

[0062] In the collision simulation, the structural parameters of the end friction energy-absorbing structure can be used as input parameters, and the vertical displacement and acceleration of each car body of the multi-coupled train can be used as outputs to study the influence of the structural parameters of the train end friction energy-absorbing structure on the pose optimization of the multi-coupled train. During the optimization process, the minimum vertical displacement and acceleration of the car body are used as the optimization objectives, and the optimal decision-making method through the Pareto solution set is used to obtain the best combination of energy-absorbing structure parameters, so that the vertical displacement and acceleration of the train reach the minimum under collision or other dynamic load conditions.

[0063] For example, the specific formula of the optimization objective is as follows:

[0064] ;

[0065] where represents the acceleration of the i-th stationary car; represents the acceleration of the j-th stationary car; represents the vertical lift of the i-th stationary car; represents the vertical lift of the j-th moving car; is the parting surface diameter of the friction tube and friction shoe for friction energy absorption; is the shoe thickness coefficient; is the pre-tightening pressure of the friction shoe; is the pre-connection length of the friction tube and friction shoe; is the dynamic friction coefficient of the friction rubber; is the taper coefficient. Preferably, during the calculation process of optimal decision-making through the Pareto solution set, the variables 、 、 and The gain weights are respectively set to 0.58, 0.12, 0.06 and 0.24. The aforementioned weight setting is based on in-depth analysis and understanding of the roles of various variables in the overall system performance. Through this weight allocation, the optimization process of the energy-absorbing structure can be better guided to achieve the best balance among key performance indicators such as strength and lightweighting.

[0066] Furthermore, in the coupled train collision dynamics model, for the coupler draft gear and anti-climbing device, the constitutive relationship of the local plastic deformation of the coupler draft gear structure is defined by a generalized spring, and the mechanical characteristic curves of the energy-absorbing structure and the coupler are constructed using a nonlinear hysteresis mathematical model.

[0067] As Figure 2 shown, the green line represents the loading curve, the red line represents the unloading curve, the abscissa X represents displacement, and the ordinate Y represents force; and the serial numbers ①~⑧ in the figure respectively correspond to the following eight parts (1)~(8) of the hysteresis behavior during the continuous loading and unloading process of the coupled train, including in sequence:

[0068] (1), Loading is carried out along the loading curve until the displacement reaches .

[0069] (2), Unloading is carried out along the hysteresis slope, and then loading is carried out again to .

[0070] (3), Loading is carried out again along the loading curve until the maximum value is reached.

[0071] (4), Unloading is carried out again along the hysteresis curve.

[0072] (5), Unloading is carried out along the unloading curve.

[0073] (6), Loading is carried out in the reverse direction along the unloading curve.

[0074] (7), Loading is carried out in the upper right direction along the hysteresis slope until the maximum value is reached.

[0075] (8), Further loading is carried out again along the loading curve.

[0076] Similarly, in the reverse direction of loading, loading and unloading will be carried out in the opposite direction.

[0077] In this embodiment, the nonlinear hysteresis mathematical model is described in the following three stages:

[0078] The calculation equation in the loading stage is: ; where is the impedance force when the coupler or anti-climbing device is loaded; is the initial pressure of the coupler or anti-climbing device; is the sliding frictional force; is the static frictional force; represents the relative longitudinal velocity between the moving end and the stationary end of the coupler and draft gear, represents the Stribeck critical judgment velocity.

[0079] The initial pressure of the coupler and anti-climbing device will affect the friction coefficient at the beginning of the impact. During the actual calculation process, it is often calculated as a constant and rewritten in the form of stiffness as: ; where, is the linear stiffness of the coupler and draft gear during loading; is the stiffness under the simultaneous action of sliding and static friction; represents the displacement of the buffer along the longitudinal direction of the train during the loading stage.

[0080] In the unloading stage, the impedance force of the coupler and draft gear and the stroke can be regarded as linearly related, and the calculation method is: ; is the linear stiffness in the unloading stage.

[0081] Among them, there are fluctuations between loading and unloading. The moving section and the stationary section of the coupler and draft gear are relatively stationary, and the relative velocity between them is 0, thus resulting in the zero-velocity stage.

[0082] When the relative velocity between the stationary end and the moving end of the coupler and draft gear system is zero, the direction of the velocity will change, resulting in non-uniqueness of the magnitude and direction of the instantaneous force in the numerical simulation; in order to accurately determine the friction damping characteristics at the zero-velocity moment, so that the magnitude of the impedance force in the calculation process has a unique solution and avoid the fluctuation phenomenon in the simulation output, the mathematical model of the instantaneous impedance force at the zero-velocity moment is established as follows:

[0083] ;

[0084] where, is the impedance force of the coupler and draft gear device at time is the impedance force of the coupler and draft gear device at time is the coupler and draft gear displacement of the coupler and draft gear device at time is the coupler and draft gear displacement of the coupler and draft gear device at time is the geometric control parameter for controlling the change of the loading and unloading boundary; is the boundary force of the coupler and draft gear characteristic curve.

[0085] Through the displacement at time and the displacement at time to determine the boundary force , when at this time, take the impedance force during the loading stage. When at this time, take the impedance force during the unloading stage.

[0086] Thereby, a force-displacement curve can be obtained based on the target model of the predicted force-displacement curve. Based on the force-displacement curve, the energy absorption characteristics of the anti-climbing device during the impact process can be obtained. And based on the above non-linear hysteresis mathematical model, it can be determined which section of the force and the corresponding position on the force-displacement curve the current situation is in according to information such as the current impact number and the displacement of the adjacent train in the multiple consecutive impacts usually generated by the formation train. Among them, during the processing of the non-linear hysteresis mathematical model, interpolation calculation can be performed according to the force-displacement curve.

[0087] Embodiment 2

[0088] Corresponding to the above embodiment, this embodiment discloses a model data processing system for a train energy absorption structure in collision simulation, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, the method of the above embodiment is implemented.

[0089] In summary, the model data processing method and system for the train energy absorption structure in collision simulation respectively disclosed in the above two embodiments of the present invention have at least the following beneficial effects:

[0090] 1. In step S2, the force-displacement curve is predicted by the force-displacement curve. The hidden variable in the middle layer is equivalent to the cipher book for sending telegrams and is shared by the encoder and the decoder. It can be regarded as the encoded value after dimensionality reduction of the high-dimensional time series data of the force-displacement curve.

[0091] 2. In step S4, both the input and output data of the model are non-time series data, and reliable results can be achieved by using common models.

[0092] 3. To analyze the energy absorption characteristics of the friction energy absorption structure under quasi-static compression, a WED-600 electro-hydraulic servo universal testing machine was used for quasi-static compression tests. During the test, the middle beam was driven at a speed of 5 mm / min to compress the hybrid friction energy absorption structure. After the test system was started, the middle beam would compress the hybrid friction energy absorption structure downward at a constant speed until the test displacement reached 120 mm, at which point the test stopped and the load was automatically unloaded. Subsequently, the corresponding total energy absorption was calculated based on the load-displacement curve obtained from the test acquisition system. To verify the accuracy of the combined target model of the present invention, the force-displacement curves and measured crashworthiness indexes obtained from multiple groups based on the combined target model of the present invention and finite element simulations were compared; it was found that the errors of the crashworthiness indexes obtained by the method of the present invention were very small, with the largest error being only 2.65%, and the overall error being less than 5%. Thus, after the combined target model of the present invention is applied to the train collision dynamics model, it can be used to evaluate the crashworthiness of trains without the participation of finite element model software and guide the parameter optimization of the energy absorption structure.

[0093] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, the present invention can have various modifications and changes. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A method for processing model data of a train energy absorption structure in collision simulation, characterized in that Including: Step S1: Obtain the force-displacement curves of different energy absorption structures based on finite element simulation to construct a first data set; Step S2: Use the force-displacement curves of each sample in the first data set as the input and output of a self-supervised network model, and extract the intermediate layer latent variables corresponding to each force-displacement curve after dimensionality reduction in the self-supervised network model; Step S3: Construct a second data set according to the intermediate layer latent variables corresponding to the structural parameters of different energy absorption structures in the self-supervised network model; Step S4: Train, validate, and test a new network model with the structural parameters of the energy absorption structure as the input and the intermediate layer latent variables as the output according to the second data set; Step S5: Combine the new network model with the decoder part of the self-supervised network model to obtain a target model for predicting the force-displacement curve based on the structural parameters, and the weights, offset value information, and processing logic of the combined decoder part are consistent with the decoder weights, offset value information, and processing logic obtained in Step S2; Step S6: Embed the target model into the coupled train collision dynamics model to predict and output the force-displacement curve required for collision simulation according to the input data of the dynamics simulation environment.

2. The method for processing model data of the train energy absorption structure in collision simulation according to claim 1, wherein In Step S1, the finite element simulation is specifically as follows: in finite element simulation software, construct and experimentally verify the finite element parametric model of the train energy absorption structure. The finite element parametric model encapsulates the entire process code of geometric parameter driving, material association, load application, mesh generation, and calculation submission, realizing the automated mapping of "parameter input - simulation output", and the first data set is constructed based on the finite element parametric model using the experimental design method.

3. The method for processing model data of the train energy absorption structure in collision simulation according to claim 2, characterized in that In Step S2, convert the force-displacement data of each sample into a 2×1000-dimensional matrix, and then use four algorithms, namely AE, RNN-AE, LSTM-AE, and GRU-AE network models, for model training. Compare and analyze the prediction accuracies of different models, and use the network model with the highest accuracy of the predicted output curve as the self-supervised network model.

4. The method for processing model data of the train energy absorption structure in collision simulation according to claim 3, characterized in that In Step S4, the new network model uses a CNN network model. The energy absorption structure is specifically the anti-climbing device in the coupler draft gear. The structural parameters include: the parting surface diameter of the friction tube and the friction shoe, the pre-tightening pressure of the friction shoe, the dynamic friction coefficient of the friction rubber, the pre-connection length between the friction tube and the friction shoe, the shoe thickness coefficient, and the taper coefficient.

5. The method for processing model data of the train energy absorption structure in collision simulation according to claim 4, characterized in that, In the coupled train collision dynamics model, the coupler draft gear includes a coupler and the anti-climbing device. The constitutive relationship of the local plastic deformation of the coupler draft gear structure is defined by a generalized spring, and the mechanical characteristic curves of the energy absorption structure and the coupler are constructed using a nonlinear hysteresis mathematical model. The nonlinear hysteresis mathematical model is described in the following three stages: The calculation equation in the loading stage is as follows: ; where is the impedance force when the coupler or anti-climbing device is loaded; is the initial pressure of the coupler or anti-climbing device; is the sliding friction force; is the static friction force; represents the relative longitudinal velocity between the moving end and the stationary end of the coupler buffer; represents the Stribeck critical judgment speed; The initial pressures of the coupler and anti-climbing device will affect the friction coefficient at the beginning of the impact. During actual calculations, they are often calculated as constants and rewritten in the form of stiffness as follows: ; where is the linear stiffness of the coupler buffer during loading; is the stiffness under the simultaneous action of sliding and static friction; represents the displacement of the buffer along the longitudinal direction of the train during the loading stage; During the unloading stage, the hook buffer resistance force and the stroke are regarded as linearly related, and the calculation method is as follows: ; is the linear stiffness during the unloading stage; Among them, there is a fluctuation between loading and unloading. The coupler draft gear moving section and the stationary section are relatively stationary, and the relative velocity between them is 0, thus resulting in the zero-velocity stage; When the static end and the moving end of the coupler buffer system are in a stage where the relative speed is zero, the direction of the speed will change, resulting in non-uniqueness of the magnitude and direction of the instantaneous force in numerical simulation; in order to accurately determine the friction damping characteristics at the zero-speed moment, so that the magnitude of the impedance force in the calculation process has a unique solution and avoid fluctuations in the simulation output, the following mathematical model of the instantaneous impedance force at the zero-speed moment is established: ； Among them, is the impedance force of the coupler draft gear at time is the impedance force of the coupler draft gear is the coupler draft displacement of the coupler draft gear at time is the coupler draft displacement of the coupler draft gear is the geometric control parameter for controlling the change of the loading and unloading boundary; is the boundary force of the coupler draft characteristic curve By Moment displacement and Moment displacement to determine the boundary force When , Take the impedance force in the loading stage. When , Take the impedance force in the unloading stage.

6. The method for processing model data of the train energy absorption structure in collision simulation according to claim 5, characterized in that, In steps S2 and S5, before model training, the force-displacement curve and the structural parameters are respectively normalized. During the normalization process, all values are directly normalized by dividing by a maximum value, and the calculation formula is: ； ； Wherein: and are the normalized value and the true value of the calculation result of the j-th group of experimental designs corresponding to the i-th input structure parameter; max is the maximum true value of the i-th input structure parameter in all the calculation results of the experimental designs; and are respectively the normalized value and the true value of the k-th point on the impact force-displacement curve in the calculation result of the j-th group of experimental designs; max is the maximum true value of the i-th crashworthiness index in all the calculation results of the experimental designs.

7. The method for processing model data of the train energy absorption structure in collision simulation according to any one of claims 1 to 5, characterized in that It also includes: In the collision simulation, the structural parameters of the end friction energy-absorbing structure are used as input parameters, and the vertical displacements and accelerations of each car body of the multi-unit train are used as outputs to study the influence of the structural parameters of the end friction energy-absorbing structure of the train on the pose optimization of the multi-unit train; and during the optimization process, the minimum vertical displacements and accelerations of the car body are used as optimization objectives, and the best combination of energy-absorbing structure parameters is obtained through the method of optimal decision-making with the Pareto solution set, so that the vertical displacements and accelerations of the train reach the minimum under collision or other dynamic load conditions.

8. The method for processing model data of the train energy absorption structure in collision simulation according to claim 7, wherein The specific formula of the optimization objective is as follows: ； In the formula, represents the acceleration of the i-th stationary vehicle; represents the acceleration of the j-th stationary vehicle; represents the vertical lift of the i-th stationary vehicle; represents the vertical lift of the j-th moving vehicle; is the parting surface diameter of the friction tube and friction shoe for friction energy absorption; is the shoe thickness coefficient; is the pre-tightening pressure of the friction shoe; is the pre-connection length of the friction tube and friction shoe; is the dynamic friction coefficient of the friction rubber; is the taper coefficient.

9. A model data processing system for a train energy absorption structure in collision simulation, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the method according to any one of claims 1 to 8 above.

Images