DHMM+SVM在切削颤振中的应用  

Application of HMM and SVM in Cutting Chatter

作　　者：蒋文凤[1] 江涌涛[1] 张春良[1] 刘琼[1] 

机构地区：[1]南华大学机械工程学院,湖南衡阳421001

出　　处：《机电产品开发与创新》2009年第3期179-181,共3页Development & Innovation of Machinery & Electrical Products

摘　　要：根据切削颤振的特点,结合隐马尔可夫模型(Hidden Markov Model,HMM)和支持向量机(Support Vector Machine,SVM)的特点,提出了一种新的状态预测技术,同时也提出了一种新的特征提取方法。首先在等时间间隔内对切削信号实时进行小波包分解,然后通过SVM对各频带区间能量变化趋势进行回归预测,最后通过HMM对预测结果进行分类。结果表明,该方法取得了较好的预测结果。A new cutting chatter system has been developed according to Hidden Markov Model （HMM） and Support Vector Machine （SVM） . This system uses HMM as the recognition method and SVM as the prediction method. Meanwhile, means like wavelet package de- composition are also employed to extract the cutting features. The basic idea and general steps are as follow. Firstly, bootstrapping analyzing the cutting signal m the same interval using wavelet packet decomposition. Secondly, we use SVK algorithm to predict the trend of energy transition. The results at last are input to HMM through scalar quantization to determine whether it is in chatter pattern. The simulation re- suits show that the new predicting method has good discriminating performances.

关 键 词：切削颤振 小波包分解 HMM SVR 

分 类 号：TG506[金属学及工艺—金属切削加工及机床]