微细电极形状损耗分类预测及控形方法分析  

Shape Wear Form Prediction and Analysis of Shape Control Method of Micro Electrode

作　　者：王慧 王元刚[1] 李晓鹏[1] WANG Hui;WANG Yuangang;LI Xiaopeng(School of Mechanical Engineering,Dalian University,Dalian 116622,China)

机构地区：[1]大连大学机械工程学院,辽宁大连116622

出　　处：《电加工与模具》2020年第5期10-13,共4页Electromachining & Mould

基　　金：国家自然科学基金资助项目(51005027)。

摘　　要：为实现微细电火花孔加工的电极形状损耗形式的分类预测,选择SVM支持向量机、BP神经网络、Logistic回归、KNN临近算法四种典型的分类算法,分别建立了微细电极损耗形式的分类模型,并根据分类情况,结合理论分析,探讨了形状控制的一般方法,同时设计了验证实验。结果表明:Logistic回归模型最为合适,能较好贴合实验数据;在小的脉冲能量下选取大的脉冲宽度有利于实现电极的均匀损耗,且此种方法具有一定的通用性。研究成果初步实现了微细电极的形状预测与控制,对提高微细电火花加工精度具有一定的指导意义。In order to realize the classification and prediction of electrode shape wear form in micro-EDM drilling,four typical classification algorithms,Support Vector Machine(SVM),BP neural network,Logistic regression and K-Nearest Neighbor,are selected to establish the classification model of micro electrode wear form.According to the classification and theoretical analysis,the general method of shape control is discussed,and the verification experiment is designed.The results show that the Logistic regression model is the most suitable,which can fit the experimental data well,and the selection of large pulse width under small pulse energy is beneficial to the realization of electrode even wear,and this method has a certain universality.The research results preliminarily realize the shape prediction and control of micro electrode,which has certain guiding significance for improving the accuracy of micro-EDM.

关 键 词：微细电火花孔加工 形状损耗 分类模型 控形方法 

分 类 号：TG661[金属学及工艺—金属切削加工及机床]