基于疲劳试验的车轮拓扑优化和多目标优化

摘要

Based on dynamic bending fatigue test and dynamic radial fatigue test,a united topology optimi-zation is performed on the disk of a-type wheel,and an assembled wheel structure with a magnesium alloy rim and an aluminum alloy disc is designed.Finite element models of wheel for dynamic bending fatigue test and dynamic ra-dial fatigue test are established to calculate the strength,stiffness,fatigue life and the safety factor of radial fatigue life,and to analyze the relationship between these performances and wheel structure. The parametric models for the assembled wheel under two working conditions are built with 12 design variables defined by using mesh deformation technique. Multi-objective optimization model for wheel is created with Isight software platform integrating softwares for the calculation of performance indicators. Both designs of experiment schemes(Hamersley and optimal Latin hy-percube) are adopted to extract 72 and 10 samples respectively to fit the Kriging surrogate model with its accuracy verified. Based on the surrogate model built, with minimizing wheel mass and maximizing the bending fatigue life and the safety factor of radial fatigue life of wheel as objectives,and stresses,displacements and flexibilities as con-straints,a multi-objective optimization of wheel is conducted with NSGA-Ⅱ and the Pareto frontiers are obtained, from which a compromised solution is selected as the optimal result with comprehensive consideration of various per-formances of wheel,and the comprehensive performances of wheel before and after optimization are compared The results show that the mass of the assembled wheel optimized is 29.42% less than that of the cast aluminum alloy wheel of same category,while meeting the requirements of the comprehensive performances of wheel.%基于动态弯曲疲劳试验和动态径向疲劳试验对16×61/2 J型车轮的轮辐进行了联合拓扑优化,设计了一个带有镁合金轮辋和铝合金轮辐的组装式车轮结构.建立其弯曲疲劳试验和径向疲劳试验的有限元模型,计算其强度、刚度、疲劳寿命和径向疲劳寿命安全系数,并分析了这些性能与车轮结构之间的关系.利用网格变形技术建立了组装式车轮在两种工况下的参数化模型并定义了12个设计变量,使用Isight软件平台集成各性能指标的计算软件建立了车轮多目标优化模型,利用哈默斯雷和最优拉丁超立方试验设计分别提取了72和10个样本点,拟合了Kriging近似模型并检验了近似模型的精度.利用所建立的近似模型,以车轮质量最小、弯曲疲劳寿命和径向疲劳寿命安全系数最大为目标,应力、位移和柔度为约束,采用第二代非劣排序遗传算法(NSGA-Ⅱ)对车轮进行了多目标优化,得到了Pareto前沿,综合考虑了车轮各项性能,选取了一个妥协解作为优化结果,并对优化前后车轮综合性能进行了对比.结果表明,在满足车轮各项性能要求的条件下,优化得到组装式车轮的质量比同型铸造铝合金车轮减小了29.42%.