基于小波网络的非导电工程陶瓷电火花加工效果预测模型  

Forecasting Model of EDM for Non-conductive Engineering Ceramics Based on Wavelet Neural Network

作　　者：于丽丽[1] 刘永红[1] 蔡宝平[1] 纪仁杰[1] 董欣[1] 

机构地区：[1]中国石油大学机电工程学院,山东东营257061

出　　处：《电加工与模具》2008年第2期25-28,共4页Electromachining & Mould

基　　金：国家自然科学基金资助项目(50675225);教育部博士点基金资助项目(20040425504)

摘　　要：针对非导电工程陶瓷双电极同步伺服电火花加工中,由于工艺参数与加工效果间的高度非线性,难以建立精确的数学模型进行效果预测的问题,提出了一种基于小波网络的建模方法。为使小波网络具有更好的学习精度和更快的收敛速度,将遗传算法和BP算法结合起来作为小波网络的学习算法。经验证基于混合学习算法的小波网络较传统神经网络模型具有更快的收敛速度和更高的预测精度。A model based on wavelet neural network to predict the process of electro-discharge machining（EDM） with synchronous servo double electrodes for non-conductive engineering ceramics is introduced due to the high nonlinear features among parameters and effects during processing. A wavelet neural network based on genetic algorithm and neural network is put forward to get higher accuracy and faster speed. The results show that the wavelet neural network model provides accurate results for process, and fits the experiments result better.

关 键 词：小波网络 遗传算法 非导电工程陶瓷 电火花加工 

分 类 号：TG661[金属学及工艺—金属切削加工及机床]