The Dynamics Simulation Of Metro Vehicle Based On Radial Basis Functions

ABSTRACT:Along with the development of urban rail transit construction in China, the requirement on the hunting stability and curving performance of metro vehicles is more and more strict. In order to solve the problem, in this paper the hunting stability and curving performance are considered as optimization objectives, the indexes of derailment coefficient, rate of wheel load reduction and wheel-rail lateral force are considered as constrains, and the suspension parameters of metro vehicles are considered as design variables. Consequently, the hunting stability and curving performance are improved obviously.Firstly, an analysis model of the hunting stability and curving performance need to be built with the parameters of metro vehicle of type B. Secondly, the sample points of metro vehicles are obtained by means of Optimal Latin hypercube design method. Thirdly using the dynamic simulation software ADAMS/Rail, the output index of critical speed, wheel-rail wear, wheel-rail lateral force, wheel load reduction rate, and derailment coefficient can be figured out in the way of entering each group of design variables. Consequently samples are obtained by combining each group of design variables and output index. And the Surrogate model of the hunting stability and curving performance can be established with the method of Radial Basis Functions. At last the results of optimization should be obtained with the method of linear weighting algorithm and NSGA-II multi-objective optimization algorithm separately based on the surrogate model. Then comparison should be carried out between optimization result and original metro vehicle.It can be find out that the primary axle positioning stiffness has a significant effect on the hunting stability and curving performance through the samples obtained from simulation. The accuracy of surrogate is correct and it can be used in optimization. From the result of optimization, it can be discovered that much more designs are presented by the NSGA-II multi-objective optimization algorithm. But the linear weighting algorithm can only obtain a result every once calculation. What’s worse, the optimization result is always affected by the weighting factor obviously.