Robust and Optimal Design of Railway Vehicle System for Derailment Risk Using Efficient Global Optimisation Method

This paper presents an innovative optimisation procedure, combining uniform design (UD) and the efficient global optimisation (EGO) algorithm, to generate a set of robust suspension parameters in a railway vehicle model. Nonlinear dynamic analysis of a 31 degree-of-freedom (DOF) railway vehicle model was determined using Kalker's linear theory and heuristic nonlinear creep criterion. To increase running safety, optimisation design for suspension parameters is introduced to make the performance more robust and reduce the sensitivity to noise. Considering the noise factors, vehicle speed and rail irregularity, the dynamic response and derailment quotient are obtained by the Runge-Kutta method. By applying uniform design (UD), Kriging interpolation and efficient global optimisation (EGO) algorithm, the best signal-to-noise ratio of the derailment quotient is increased from 12.05 dB to 31.3 dB, or 160%. The numerical results indicate that the optimal and robust design of suspension parameters has been determined successfully by the novel optimisation process.