Search

The multi-objective optimisation of high-speed rail profile with small radius curve is studied in the paper. A multi-objective mathematical model for rail profile optimisation of high-speed railway is established. The CN60 rail profile is parameterised into a series of generalised functions of design variables. In order to guarantee the smoothness of the rail profile and meet the maximum grinding depth of rail in China, the constraints are employed. The wheel-rail vertical clearance and equivalent conicity of wheelset are taken as objective functions, and a rail wear prediction programme is compiled. Contact line method is employed to complete the detection algorithm of wheel-rail contact points. Finally, NSGA-II Algorithm is adopted to solve the Pareto-optimal front of the optimisation model. A set of solutions are retrieved from the Pareto optimal front solution as the optimised profile. The optimised rail profile and the original rail profile are matched with the LMA wheel profile (a certain worn type of wheel profiles for EMU in China) respectively. It is testified that the rail profile could effectively reduce the rail wear and improve curving performance. The new method proposed in this paper can provide some reference for the optimisation design of high-speed rail profile with small radius curve. [ABSTRACT FROM AUTHOR]

Ride comfort indices are used to evaluate the vibration of rail vehicles and to measure the discomfort of passengers. Among the various ride comfort indices, the Sperling index is widely used in China and other countries. However, there are several different methods for determining the Sperling index from dynamic simulations and performance measurements of rail vehicles, and the results obtained by different algorithms are inconsistent. It is therefore difficult to make an accurate evaluation with the different calculation results. In this paper, a comparison is made between algorithms based on time domain and frequency domain analysis and using the second and third powers of the acceleration. The different algorithms are then summarised into a unified equation, and the consistency of the results is analysed by this equation. It is found that only the r.m.s-based algorithm in the time or frequency domain is stable when analysis is carried out over different sample times. The time-frequency consistency of the r.m.s-based algorithms is verified by Parseval's theorem and using calculation results from time domain simulations. [ABSTRACT FROM AUTHOR]