瑞风商务车托架拉延成形数值模拟及工艺参数优化  

Numerical Simulation and Process Optimization of Stretching Forming for Bracket of Refine Business Vehicles

作　　者：李萍[1] 薛克敏[1] 杨文江[2] 

机构地区：[1]合肥工业大学材料科学与工程学院,合肥230009 [2]江淮汽车股份有限公司,合肥230022

出　　处：《机械工程学报》2008年第7期176-180,共5页Journal of Mechanical Engineering

基　　金：国家自然科学基金(50405020);江淮汽车-合肥工大汽车技术研究院科技资助项目

摘　　要：基于逆向工程建立瑞风商务车托架零件的几何模型,并基于Dlymform软件平台对不同工艺参数下该零件的拉延成形过程进行数值模拟。在此基础上,以压边力、拉延筋高度和拉延筋圆角半径作为设计变量,以零件不发生破裂为优化目标,以有限元数值模拟结果作为虚拟样本,建立目标函数的人工神经网络预测模型;将人工神经网络预测模型作为优化算法的知识源,采用遗传算法对压边力、拉深筋几何参数等工艺参数进行了优化设计。试验结果表明,数值模拟、神经网络预测和工艺优化是可靠的,从而可为制定金属板料最佳的冲压成形工艺提供一条先进、合理的途径。Based on reverse engineering, geometric model of bracket of Refine business vehicles is reconstructed. By using Dynaform software, the stretching forming of the part is simulated with different process parameters. Based on the Dynaform software platform, the prediction model of object function is established by using artificial neural network and is regarded as knowledge source of optimization algorithm. In object function, blank holder force, draw bead height and fillet radius are design variables and prevention of cracking is considered as the optimization objective. The virtual training samples of network are obtained by finite element simulating results. Process parameters optimization is performed with genetic algorithm. The experimental results indicate that the numerical simulation is effective and the process optimization based on artificial neural network and genetic algorithm is feasible. An effective mean is offered for determining optimum deformation process parameters of sheet metal forming.

关 键 词：拉延 数值模拟 工艺优化 人工神经网络 覆盖件 

分 类 号：TG302[金属学及工艺—金属压力加工]