Tool Wear Condition Monitoring based on Principal Component Analysis and CSupport Vector Machine

Abstract:

In order to monitor tool wear condition（TWC）, the powersensorbased monitoring system on the state of machining tool wear was designed. The monitoring model of TWC was proposed based on principal component analysis（PCA）and Csupport vector machine（CSVM）. Current and power signals were obtained from power sensor during cutting process. After that, the features of these signals were extracted using PCA. The principal components，mainly affecting TWC, were chosen as the input samples of CSVM to carry out monitoring the tool condition with accuracy. The results of computerized numerical control（CNC） turning machine tool show that the model is effective even in the case of a small samples. Moreover, a comparison about the monitoring and prognostics capability between the presented method and back propagation（BP）neural network has been made.