文献类型：期刊文献

中文题名：基于Dynaform与RBF-NSGA-II算法的冲压成形工艺参数多目标优化

英文题名：Multi-objective optimization of stamping forming process parameters based on Dynaform and RBF-NSGAII algorithm

作者：刘强[1,2,3];俞国燕[1,2,3];梅端[4]

机构：[1]广东海洋大学机械与动力工程学院,广东湛江524088;[2]南方海洋科学与工程广东省实验室(湛江),广东湛江524088;[3]广东省海洋装备及制造工程技术研究中心,广东湛江524088;[4]广东海洋大学数学与计算机学院,广东湛江524088

年份：2020

卷号：27

期号：3

起止页码：16

中文期刊名：塑性工程学报

外文期刊名：Journal of Plasticity Engineering

收录：CSTPCD、、北大核心2017、Scopus、CSCD2019_2020、北大核心、CSCD

基金：广东省普通高校重点科研项目(2018KZDXM038);南方海洋科学与工程广东省实验室(湛江)资助项目(ZJW-2019-01);广东海洋大学科研启动资助项目(R19016);湛江市科技项目(2018A01019,2017A03005,2019B01004,2019B01006)。

语种：中文

中文关键词：多目标优化;冲压成形;田口方法;Dynaform;径向基函数;带精英策略的非支配排序遗传算法;逼近理想解排序法

外文关键词：multi-objective optimization;stamping forming;Taguchi method;Dynaform;RBF;NSGA-Ⅱ with the elitist strategy;TOPSIS

中文摘要：为了解决冲压成形工艺多目标优化问题,提出一种基于Dynaform与智能算法融合的多目标优化方法。以空调压缩机壳冲压成形为例,以减小最大减薄率和最大增厚率为优化目标,建立CAD模型并利用有限元分析软件Dynaform进行冲压成形数值仿真,映射其物理过程;采用田口正交试验法进行冲压成形仿真试验安排,并对试验结果进行极差和方差分析,综合评估冲压过程中冲压速度、摩擦系数、压边力、板料厚度和模具间隙对冲压成形质量的影响程度和影响规律;利用仿真试验数据训练径向基函数(RBF)神经网络,结合带精英策略的非支配排序遗传算法(NSGA-Ⅱ)获得Pareto最优解集,进而通过逼近理想解排序法(TOPSIS)评价筛选出最优工艺。利用Dynaform对所得最优工艺进行有限元分析,验证方法的有效性。

外文摘要：To solve the problem of multi-objective optimization in stamping process,a multi-objective optimiztion method based on Dynaform and intelligent algorithm was proposed.Taking the stamping forming of the air conditioner compressor shell as the example,reducing the maximum thinning rate and the maximum thickening rate were taken as the optimization goal,the CAD model was established and the finite element analysis software Dynaform was used to carry out the numerical simultion of stamping forming to map the physical process.Taguchi orthogonal test method was used to arrange stamping forming simulation test,and the range and variance of the test results were analyzed to comprehensively evaluate the effect degree and law of stamping speed,friction coefficient,blank holder force,sheet thickness and die gap on stamping forming quality during the stamping process.The radial basis function(RBF)neural network was trained with the simulation experimental data,and Pareto's optimal solution set was obtained by RBF coupled with the non-dominant sorting genetic algorithm Ⅱ(NSGA-Ⅱ)with the elitist strategy,and then the optimal process was evaluated and screened by technique for order preference by similarity to ideal solution(TOPSIS).The optimal process was analyzed by Dynaform to verify the effectiveness of the method.