摘要
基于人工神经网络具有良好的非线性逼近特性,以先期人工经验为依据,利用正交试验结果作为神经网络的训练样本,建立改进的Vogl快速BP网络模型,并通过样本检测模型的预测准确度,在正交试验最佳组合处采用小步长搜索方法,结合神经网络模型进一步优化工艺参数.结果表明:结合正交试验,神经网络和先期人工经验进行球面磨削工艺参数优化可以缩短参数优化时间,提高工艺设计效率,改善表面加工质量.
According to previous artificial experience, the data from an orthogonal experiment was used as the training sample to establish a neural network model on the basis of neural network＇s non-linear approach characteristic. An improved fast vogl BP neural network was established and the precision of mode was forecasted through the sample. Then small-step method was used with the best parameter got from orthogonal experiment and technical parameters were further bettered. The result shows that the integration of orthogonal experiment ,ANN and previous artificial experience for sphere grinding greatly increases the technological design efficiency and improves the surface quality.
出处
《上海交通大学学报》
EI
CAS
CSCD
北大核心
2009年第12期1956-1961,共6页
Journal of Shanghai Jiaotong University
基金
上海市重大技术装备研制专项基金资助项目(0706014)
关键词
正交试验
神经网络
球面磨削
球面粗糙度
orthogonal experiment
neural network
sphere grinding
sphere roughness
