采用代理模型的汽车铝合金轮毂优化设计

Optimization Method of Automobile Aluminum Alloy Wheel Hub Based on Surrogate Model

提出基于神经网络代理模型的汽车铝合金轮毂优化方法。对铝合金轮毂的两种工况开展静力学分析,结果表明相比正对辐条加载,正对窗口加载时轮毂的受力状态更加恶劣,不同加载方式应力都集中于外侧轮辐与辐板连接处附近。设计并实施静力学试验,验证有限元分析的准确性。基于静力学分祈结果,提出分.层递进的轮轂轮辐优化方法,通过正交试验设计神经网络代理构建和遗传算法寻优,实施轮辐尺寸优化、轮辐截面形状优化、轮辐圆角优化,最后对结构进行加强优化。与最初模型相比,优化后轮毂重量下降1.91%,正对辐条加载位移下降11.01%,最大应力下降7.54%,正对窗口加载最大位移下降10.79%,最大应力下降20.79%,轮毂力学性能有了较大幅度提高,同时达到了轻量化的目的。本文提出的优化策略可为轮毂结构优化提供有益参考。

An optimization method of automobile aluminum alloy wheel hub based on neural network surrogate model is proposed.The static analyses of the two working conditions of the aluminum alloy wheel hub are carried out.Results show that the condition of the wheel hub under window load is worse than that under spoke load,and the stress is concentrated near the connection between the outer spoke and the spoke plate.The static experiment of the wheel hub is designed and implemented to verify the accuracy of the finite element analysis.Based on the results of static analysis,a layered and progressive wheel hub spoke optimization method is proposed,consisting of orthogonal experimental design,neural network surrogate model construetion and genetic algorithm search for optimization,which is implemented in spoke size optimization,spoke cross section shape optimization,and spoke fllt optimization.Finally,the wheel hub with optimized structure is oblained.Compared with the original model,the weight of the optimized wheel hub,the maximum displacement under spoke load,the maximum stress,the maximum displacement under the window load and the maximum stress are reduced by 1.91%,11.01%,7.54%,10.79%and 20.79%,respectively.The mechanical performance of the wheel hub has been efctively improved,while achieving the purpose of lightweight.The proposed optimization can provide a useful reference for the optimization of the wheel hub structure.