Estimation of Vehicle Dynamic Response from Track Irregularity Using Deep Learning Techniques

Abstract

To improve the quality of track maintenance work, it is a desire to estimate vehicle dynamic behavior from track geometry irregularities. This paper proposes a deep learning model to predict vehicle responses (e.g., vertical wheel-rail forces, wheel unloading rate, and car body vertical acceleration) using deep learning techniques. In the proposed CA-CNN-MUSE model, convolutional neural networks (CNNs) are used to learn features of track irregularities, and multiscale self-attention mechanisms (MUSE) are employed to capture the long-term and short-term trends of sequences. Coordinate attention (CA) is introduced into CNN to focus on important interchannel relationships and important spatial mileage points. The experiments were performed on a multibody simulation model of the vehicle system and the measured data of the actual high-speed line. The results show that the CA-CNN-MUSE has high prediction accuracy for vertical vehicle responses and fast computation speed. The predicted time-domain waveforms and power spectral densities (PSDs) agree well with the actual vehicle responses. The main features of the lateral vehicle responses can also be captured by the proposed method, yet the results are not as good as the vertical ones.

Vibration characteristics of defective axle box bearings in high-speed trains under track irregularity excitation

“…However, during the actual operation of high-speed trains, the axle box bearings of high-speed trains are coupled with other components, which will be subject to various external excitations, among them, wheel-rail excitation is the main source of vibration acceleration experienced by various components of the train. Such as wheel polygon [28][29][30][31][32], wheel flat scar [33][34][35], track irregularity [2,33,36,41], etc. Based on the multi-body dynamics theory, the literature [29] established a rigid-flexible coupling dynamic model of vehicle-track, analyzed the influence of polygon amplitude on the time-frequency domain of wheel noise, and the research results showed that when the polygon order is 20, with the increase of polygon amplitude, the wheel-rail vertical force and the acceleration of wheel-rail and rail plate gradually increased.…”

“…In recent years, deep learning-related technologies have flourished [37][38][39][40] and have gradually been applied to the field of model analysis. The literature [41] proposed a deep learning model that uses deep learning technology to predict the vehicle response under track irregularities. The model successfully simulated the vertical wheel-rail force and body acceleration, but the estimation of lateral wheel-rail force and body acceleration is not as good as the results for vertical wheel-rail force.…”