摘要
为了提高滚刀刀圈的轧制产品质量,提出了基于家族遗传算法的轧制工艺参数优化方法。介绍了滚刀刀圈的轧制成形方法和原理,减小滚刀刀圈的等效应力和等效应变为目标建立了优化模型。使用随机抽样法和Deform-3D有限元仿真得到了样本数据,使用自适应学习神经网络对工艺参数和质量参数进行了模型回归。经验证,自适应学习神经网络的拟合效率与预测精度高于传统神经网络。在遗传算法中引入家族并行进化和族间交叉策略,提高算法优化能力。经验证,家族遗传算法优化的适应度函数值小于传统神经网络,且获得最优值的迭代次数更少。经仿真和实验验证可知,优化后的轧制件外观合格,且硬度整体高于优化前,验证了刀圈轧制优化方法的有效性。
In order to improve the rolling quality of cutter ring,an optimization method of rolling process parameters based on race genetic algorithm was proposed. The rolling forming method and principle of hob ring are introduced. The optimization model is established to reduce the equivalent stress and equivalent effect of hob ring. The random sampling method and DEFORM-3D finite element simulation are used to obtain the sample data,and the adaptive learning neural network is used to model regression of process parameters and quality parameters. It is verified that the fitting efficiency and prediction accuracy of the adaptive learning neural network are higher than that of the traditional neural network. Race parallel evolution and cross strategy are introduced to improve the optimization ability of genetic algorithm. It is verified that the fitness function optimized by the family genetic algorithm is smaller than that of the traditional neural network,and the number of iterations to obtain the optimal value is less. The simulation and experimental results show that the appearance of the optimized parts is qualified,and the hardness is higher than that before optimization,which verifies the effectiveness of the optimization method.
作者
李奕晓
杨善国
王旭荣
LI Yi-xiao;YANG Shan-guo;WANG Xu-rong(Department of Mechanical and Electrical Engineering,Yongcheng Vocational College,He'nan Yongcheng 476600,China;Department of Mechanical and Electrical Engineering,China University of Mining and Technology,Jiangsu Xu-zhou 221004,China;Shandong Transport Vocational College,Shandong Weifang 261206,China)
出处
《机械设计与制造》
北大核心
2023年第3期82-87,共6页
Machinery Design & Manufacture
基金
山东省教育科学“十三五”规划重点资助项目(BZD2017002)。
关键词
滚刀刀圈
轧制优化
家族遗传算法
自适应学习神经网络
族间交叉
Cutter Ring
Rolling Optimization
Race Genetic Algorithm
Learning Adaptive Neutral Network
Race Crossover
