汽车发动机罩的神经网络-强繁殖NSGA-Ⅱ算法冲压参数优化  

Parameter optimization on stamping of neutral network-strong reproduction NSGA-Ⅱalgorithm for automobile engine hood

作　　者：王慧怡[1] 王岫鑫 刘学 Wang Huiyi;Wang Xiuxin;Liu Xue(Quality Control Department,Changchun Automobile Industry Institute,Changchun 130013,China;School of Bioinformatics,Chongqing University of Posts and Telecommunications,Chongqing 400065,China;College of Automobile Engineering,Changchun Automobile Industry Institute,Changchun 130013,China)

机构地区：[1]长春汽车工业高等专科学校质量控制处,吉林长春130013 [2]重庆邮电大学生物信息学院,重庆400065 [3]长春汽车工业高等专科学校汽车工程学院,吉林长春130013

出　　处：《锻压技术》2022年第7期100-106,共7页Forging & Stamping Technology

基　　金：吉林省职业教育与成人教育教学改革研究课题(2020ZCY205)。

摘　　要：为了提高车辆发动机罩内板的冲压质量,以减小冲压制件的最大减薄率和最大增厚率为目标,提出了基于神经网络-强繁殖NSGA-Ⅱ算法的冲压参数优化方法。建立了减小最大减薄率和最大增厚率的多目标优化模型。使用最优拉丁抽样法在思维空间抽取了采样点,依据数值模拟获得了采样点的性能参数。使用BP神经网络拟合冲压参数与质量参数的关系,经验证,回归精度较高,BP神经网络可以用于质量参数的预测。定义了多点随机交叉和排交叉位随机变异法,将其应用于NSGA-Ⅱ算法,给出了基于强繁殖NSGA-Ⅱ算法的优化模型求解方法。经验证,强繁殖NSGA-Ⅱ算法的Pareto解集可以支配NSGA-Ⅱ算法解集,验证了改进策略的有效性。优化后最大减薄率均值和最大增厚率均值分别减小了15.14%和18.93%,验证了优化方法的有效性和优越性。In order to improve the stamping quality of automobile engine hood inner panel and reduce the maximum thinning rate and the maximum thickening rate of stamping parts,a stamping parameter optimization method based on neural network-strong propagation NSGAⅡalgorithm was proposed,and a multi-objective optimization model for reducing the maximum thinning rate and the maximum thickening rate was established.Then,the sampling points in the thinking space were extracted by using the optimal Latin sampling method,and the performance parameters of the sampling points were obtained according to the numerical simulation.Furthermore,through using BP neural network to fit the relationship between stamping parameters and quality parameters,it was verified that the regression accuracy was high,and the BP neural network could be used to predict quality parameters.Finally,the multi-point random crossover and row crossover random mutation methods were defined and applied to NSGA-Ⅱalgorithm,and the solution method of optimized model based on strong reproduction NSGA-Ⅱalgorithm was given.The verification results show that the Pareto solution set of strong reproduction NSGA-Ⅱalgorithm can dominate the solution set of NSGA-Ⅱalgorithm,which verifies the effectiveness of the improved strategy.After optimization,the average values of the maximum thinning rate and the maximum thickening rate are reduced by 15.14%and 18.93%respectively,which verifies the effectiveness and superiority of the optimized method.

关 键 词：发动机罩内板 冲压 BP神经网络 强繁殖NSGA-Ⅱ算法 最大减薄率 最大增厚率 

分 类 号：TP319[自动化与计算机技术—计算机软件与理论]