基于正交试验、BP神经网络和遗传算法的冷挤压模具优化设计方法  

Optimization of cold extrusion die based on orthogonal experiment,BP neural network and genetic algorithm

作　　者：杨庆华[1] 占伟涛 吴海伟[1] 王志恒[1] 鲍官军[1] 

机构地区：[1]浙江工业大学特种装备制造与先进加工技术教育部重点实验室,浙江杭州310014

出　　处：《浙江工业大学学报》2015年第3期251-256,共6页Journal of Zhejiang University of Technology

基　　金：国家自然科学基金资助项目(51275475);浙江省科技计划公益技术研究工业项目(2008C21126)

摘　　要：以桑塔纳L45449汽车轮毂轴承内圈为例,建立汽车轮毂轴承内圈冷挤压模具简化模型,运用Deform-3D软件对零件成形过程进行仿真,对挤压过程中的行程载荷和模具磨损进行分析,并据此给出模具优化设计约束条件和优化目标,提出了一套结合正交实验法、BP人工神经网络和遗传算法的模具优化设计方法,对冷挤压汽车轮毂轴承内圈模具进行优化,并对优化结果进行有限元验证.结果表明:优化结果与仿真分析结果相近,最大行程载荷相对误差为4.55%,凹模磨损量绝对误差为0.06μm,提出的优化设计方法能有效缩短模具设计周期,降低模具制造成本.Taking Santana L45449 automotive wheel bearing rings for example, simulation based on finite element method for cold extrusion process was performed after the simplified model for cold extrusion die of wheel hub bearing rings was established based on the platform of Deform-3D and optimization variables and optimization objective were proposed based on the simulation results. Then a collaborative optimization method for die design was described, combined using orthogonal experiment, neural network and genetic algorithm. Cold extrusion die of wheel hub bearing rings were optimized and compared with Finite element analysis results. The results show that. the optimization results and simulation results are similar, the relative error of the maximum stroke load is 4. 55%, and the absolute error of the die wear is 0. 06 μm, so the proposed optimization method can effectively reduce the die design cycle and costs.

关 键 词：冷挤压 轴承内圈 有限元 正交实验 遗传算法 

分 类 号：TG376.3[金属学及工艺—金属压力加工]