双面渐进成形工艺参数优化及减薄率的预测  

Optimization of process parameters and prediction of thinning rate for double-sided incremental forming

作　　者：张澧桐 田雨 顾鹏 张鑫 ZHANG Litong;TIAN Yu;GU Peng;ZHANG Xin(School of Mechanical and Electrical Engineering,Changchun University of Science and Technology,Changchun 130022,CHN;Changchun Equipment Technology Research Institute,Changchun 130000,CHN)

机构地区：[1]长春理工大学机电工程学院,吉林长春130022 [2]长春设备工艺研究所,吉林长春130000

出　　处：《制造技术与机床》2024年第7期131-138,共8页Manufacturing Technology & Machine Tool

基　　金：吉林省科技发展计划项目重点研发(JJKH20220732KJ)。

摘　　要：渐进成形的减薄率是衡量成形件质量的重要指标。文章采用Box-Behnken设计实验方案进行试验,分析了刀具直径D、层间距Δz、成形角α和板厚t对减薄率的影响,并得到试验最优的参数组合。建立了工艺参数到减薄率的BP神经网络模型,用数据训练集训练网络,计算测试集减薄率预测模型的精度。针对BP神经网络平均误差大(6.42%)的问题,用粒子群算法(PSO)优化了BP神经网络模型参数,使预测误差降低到2.24%。PSO-BP神经网络模型可以有效预测工艺参数和减薄率的关系。The rate of thinning in incremental forming is a crucial indicator for assessing the quality of formed parts.In this study,we conducted experiments using a Box-Behnken design experimental scheme to analyze the impact of tool diameter(D),layer spacing(Δz),forming angle(α),and plate thickness(t) on the thinning rate.By obtaining an optimal combination of these parameters,we established a BP neural network model that correlates process parameters with thinning rate.The model was trained using a data training set and its accuracy in predicting the thinning rate for a test set was evaluated.To address the issue of high average error in the BP neural network model(6.42%),we employed particle swarm optimization(PSO) to optimize its parameters,resulting in a reduced prediction error of 2.24%.The PSO-BP neural network model effectively predicts the relationship between process parameters and thinning rate.

关 键 词：双面渐进成形 减薄率 智能神经网络 粒子群算法 正交试验 

分 类 号：TG306[金属学及工艺—金属压力加工]