细化Kriging模型在轻轨车轴优化设计中的应用

基于仿真试验,根据低地板轻轨车轴的结构特点和性能指标,将Kriging插值和拉丁超立方抽样试验相结合来构造车轴多目标优化的代理数学模型,并通过多目标遗传算法寻找该代理模型的最优解作为车轴参数的设计依据。为了使所构造的Kriging响应面能准确高效地代表稳健优化目标和约束函数,将Kriging模型在响应面精度检验时的样本点和多目标遗传算法所产生的最优解检验点进行更新,直到达到精度要求为止,从而保证所建立的响应面模型具有较高的全局拟合精度和最优解拟合精度。

Application of Refined Kriging Model in Optimization Design of Light Rail Axles

The proxy mathematical model of multi-objective optimization on axle was established by combining with Kriging interpolation and Latin hypercube sampling test based on the simulation test,which according to the structural characteristics and performance index of the low-floor light rail axle. While the optimal solution of this proxy model was sought as the design basis of axle parameters by the multi-objective genetic algorithm. Furthermore,the sample points from the response surface precision test of Kriging model and the optimal checkpoints produced by multi-objective genetic algorithm were updated to achieve the accuracy requirements,which enables the established Kriging response surface represents the robust optimization objective and constraint function accurately as well as efficiently,thus the high global fitting accuracy and optimal solution fitting accuracy of the established response surface model were realized.