A Data-Driven Dynamics Simulation Model for Railway Vehicles Based on Lightweight 3DCNN With Physics-Informed Constraints

The dynamics simulation of complex railway vehicles requires a dedicated vehicle model, such as multi-body dynamics model. However, the multi-body model is time-consuming in long-distance simulation due to its computational complexity. This issue can be alleviated by using a data-driven vehicle dynamics model due to its effective generalization and computational speed. Firstly, the construction of the physical model of the vehicle system is carried out to obtain the coupling relationship between the components. Secondly, the coupling relationship between the components is embedded into the loss function of the deep neural network as physics-informed constraints. Further, the network parameters satisfying certain physical laws are obtained by minimizing the loss function. Finally, the proposed lightweight 3D convolutional neural network is used to predict the vibration state of the vehicle system. The dynamic response resulting from both the data-driven simulation model and the multi-body simulation model are investigated and compared. The simulation results show that the data-driven dynamics simulation model can accurately predict the vibration state of the vehicle system. The data-driven simulation model has smaller size and faster operation speed, which can be applied to long-distance prediction research of vehicle systems.

I. Introduction

The vehicle-track coupling system is a complex dynamic system with strong coupling, multi-parameter, nonlinear, time-varying parameters, and uncertain factors. In order to realize the analysis and evaluation of vehicle operation status, it is the key to build a high-precision mapping model of vehicle system. At present, vehicle system modeling methods can be divided into mechanism modeling and data-driven modeling. In mechanism modeling [1], [2], [3], due to the complex wheel-rail spatial coupling relationship, the calculation of wheel-rail contact force and creep force in the vehicle-track coupling system, the long-distance dynamic simulation is extremely time-consuming and expensive. However, the long-distance dynamic simulation is an indispensable part in the actual research process, such as the research of wheel polygon wear, fatigue life of high-speed railway axles, and other fields. Cai et al. [4], [5], in order to analyze the trend of polygon wear, established a vehicle-track coupling dynamics model for polygon wear, and the evolution law of wheel polygon was simulated respectively when the vehicle traveled 550km and 33000km. Wu et al. [6] established a multi-body dynamic model of railway vehicles. The model considers the gearbox shell and wheelset as flexible, and the dynamic stress change of the gearbox with a running mileage of 140000km is simulated. Hu et al. [7] simulated the fatigue load spectrum of the Beijing-Tianjin line for about 30 minutes through multi-body dynamics simulation, which is used for the fatigue damage and residual life estimation of axles. For similar research fields, the calculation time of the simulation model established by the vehicle internal mechanism is as long as ten hours or even several days, which limits the research in related fields to a certain extent.