Model-based and Data-driven Mapping of High Speed Railway Turnout Irregularity and Vehicle Response
Track irregularity is the main factor affecting the response of high speed railway vehicles.In this paper,a vehicle-turnout-lower base rigid-flexible coupling dynamics model and a bidirectional short-time and long-time neural network model based on Bayesian optimisation(BO-BiLSTM)was established to reveal the mapping relationship between high speed railway turnout irregularity and vehicle response from the aspects of mechanistic modelling and data-driven,respectively.The results show that the short-wavelength(3～5 m)longitudinal levels and cross level in the turnout can significantly affect the wheel-rail vertical action when the train crosses the turnout,resulting in the surge of wheel-rail vertical force and wheel-weight reduction rate,and the train has a greater risk of derailment safety.The BO-BiLSTM model can achieve accurate estimation of irregularities above 2 m in wavelength for longitudinal levels,above 3 m in wavelength for alignments and gauge irregularities,and above 1.5 m in wavelength for twist.
NETL
NSTL
万方数据
