Abstract
The monitoring of cutting states is indispensable for reliable and trouble-free machining. In this paper a monitoring system for classifying the levels of tool wear in high speed machining into some categories has been developed using an ART2,one of unsupervised and self-organizing artificial neural networks. The input pattern used for the ART2 was an array of normalized wavelet coefficients of feed force. The outputs of ART2 were four or five categorized tool wear levels : the incipient stage, intermediate stage, final stage and hazardous stage in the case of four categories. For two apparently different series of the force data obtained under the same cutting conditions, which are often seen in experiment, the ART2 neural network showed quite similar classification of tool wear levels from the beginning to the end of cutting. Further study proved that this monitoring system worked well for different cutting speeds V=5-7 m/s or different feed rates &fonf;=0.10-0.20 mm/rev.
