Sensor fusion for real-time condition monitoring of tool wear in surfacing with fly cutters

Abstract

A coherent artificial neural network, ANN, software program capable of real time analysis and decision-making is utilized in this work for the automatic detection and diagnostics of tool wear during a surfacing milling operation using a fly cutter. Several sensors were utilized to collect data indirectly related to wear: current measurements from the spindle and two (x, y) drive motors, three (x, y, z) components of cutting force, and acoustic emission. Furthermore, direct wear measurements were collected using image capturing and dimensional measurements of the worn location (not performed in real-time). As the inputs from these sensors were ‘fused’, the ANN utilized this multiple-sensor data to yield reasonable predictions of ‘good’, ‘used’, and ‘worn’ tools.