Implementation of remote monitoring system for prediction of tool wear and failure using ART2

Abstract

Remote monitoring of tools for prediction of tool wear in cutting processes was considered, and a method of implementation of a remote-monitoring system previously developed was proposed. Sensor signals were received and tool wear was predicted in the local system using an ART2 algorithm, while the monitoring result was transferred to the remote system via internet. The monitoring system was installed at an on-site machine tool for monitoring three kinds of tools cutting titanium alloys, and the tool wear was evaluated on the basis of vigilances, similarities between vibration signals received and the normal patterns previously trained. A number of experiments were carried out to evaluate the performance of the proposed system, and the results show that the wears of finishing-cut tools are successfully detected when the moving average vigilance becomes lower than the critical vigilance, thus the appropriate tool replacement time is notified before the breakage.