2种响应面方法在车轮踏面优化中的应用分析比较  

Application Analysis Comparison of Two Response Surface Methods for the Optimization of the Wheel Tread

作　　者：成棣[1] 王成国[1] 刘金朝[2] 陈波[1] 

机构地区：[1]中国铁道科学研究院铁道科学技术研究发展中心,北京100081 [2]中国铁道科学研究院基础设施检测研究所,北京100081

出　　处：《中国铁道科学》2010年第3期64-69,共6页China Railway Science

基　　金：国家"八六三"计划项目(2009BAG12A01-B08-2);铁道部科技研究开发计划项目(2008J008-B)

摘　　要：首次将响应面方法应用到车轮踏面优化中。分别基于多项式响应面车轮踏面优化方法和高斯径向基函数响应面车轮踏面优化方法,用C++语言编写优化软件模块,实现在C++环境中调用多体动力学软件ADAMS/Rail进行轨道车辆系统的动力学仿真分析;将动力学仿真分析得到的数据反馈给车轮踏面优化软件模块,完成整个优化设计的循环过程。分析比较这2种响应面方法在车轮踏面优化中应用的结果表明,2种响应面方法非常适合于车轮踏面优化,而且收敛速度快;从磨耗指数的优化来看,用2种响应面方法对200 km.h-1客车车轮踏面进行优化后,新车轮踏面较原车轮踏面的磨耗指数降低52%左右,说明对磨耗指数的优化非常有效。在优化计算时间方面,多项式响应面车轮踏面优化方法为50 566 s,高斯径向基函数响应面车轮踏面优化方法为16 449 s;在1次试验设计的试验次数方面,多项式响应面车轮踏面优化方法为77次,而高斯径向基函数响应面车轮踏面优化方法只有25次。在车轮踏面优化中高斯径向基函数响应面方法优于多项式响应面方法。The response surface method was applied to the optimization of the wheel tread for the first time.The two optimization software templates,respectively based on polynomial response surface wheel tread optimization method and Gauss radial basis function response surface wheel tread optimization method,were programmed in C＋＋ language.Both optimization software templates were able to call the multibody dynamics software ADAMS/Rail to analyze dynamic simulation of railway vehicle system in C＋＋ environment and then feed the useful data from the dynamic simulation analysis to the wheel tread optimization software,to complete the cycle process of the whole optimization design.The results of applying the two response surface methods to wheel tread optimization show that these two response surface methods are rightly suitable for the wheel tread optimization with rapid convergence.In terms of the optimization of the wear index,the two response surface methods can cut the wear index of the wheel tread of 200 km·h^-1 passenger car by around 52% through wheel tread optimization,compared with that of the original wheel tread,a considerable drop of wear index which shows a quite efficient optimization of wear index.In the aspect of the optimization of time consuming,polynomial response surface wheel tread optimization method is 50 566 s and Gauss radial basic function response surface wheel tread optimization method is 16 449 s.In the aspect of times of experiment per experiment design,the polynomial response surface wheel tread optimization method is 77 experiments and the Gauss radial basic function response surface wheel tread optimization method is only 25 experiments.In the wheel tread optimization,Gauss radial basic function response surface wheel tread optimization method is superior to polynomial response surface wheel tread optimization method.

关 键 词：车轮踏面 优化方法 多项式 径向基函数 响应面 正交设计 

分 类 号：U260.331.1[机械工程—车辆工程] U270.331.1[交通运输工程—载运工具运用工程]