Abstract: The feature of dynamic cutting force and vibration signal are obtained during thecutting process,and then the wearing state are examined by statistical pattern recognition method.By time series analysis,we build models depicting the cutting tool states,coacervate information from dynamic date and construct feature vectors for discrimination.Three algorithms based on the nearest neighbour rule are used to design a classifier.They are editing technique boundary extracting and boundary patching。By using acondensed sample set processed with such algorithms,its recognition rate is near tothe result of 1-NNR with large sample set,but the speed has been increased about 6times.It seems that the method could be used to monitor on line the state of cuttingtool wearing.