摘要
盒形件成形的研究是复杂形状零件成形研究的重要基础, 本文利用人工神经网络构建了方盒件成形的目标函数模型, 将14个参数作为输入向量, 通过正交试验法对模型的训练, 得到了高精度的神经网络模型, 用于预测方盒件成形时的误差小于5%。为预测和优化方盒件拉深成形质量提供了行之有效的手段。
The study of rectangle-box parts forming is an important basis of the study of complex-shape parts forming. The paper establishes the objective functional model of square-box parts forming with ANN, using fourteen parameters as input vectors. It obtains neural network model of high precision through the training of cross experiment method. The error of forecasting square-box parts forming is less than five percent. It will offer effective method for forecasting and optimizing square-box parts forming quality.
出处
《塑性工程学报》
EI
CAS
CSCD
北大核心
2005年第1期20-23,共4页
Journal of Plasticity Engineering
关键词
方盒件
冲压成形
神经网络
正交试验法
square-box parts
press forming
neural network
cross experiment method