Multi Sensor Tool Condition Monitoring System Using Mahalanobis I-kaz Prediction System

Abstract

Problems related to cutting insert wear or tool failure during the machining process is estimated at 20%. Cutting insert wear will damage the quality, the surface and the size accuracy of the workpiece. The main purpose of TCM is to identify the cutting insert wear state at any time and to stop the machining process before the cutting insert wear exceeds the permitted limit. This paper discuses the development of a tool wear monitoring system in turning machining process using Mahalanobis I-kaz System (MIS) prediction method. A total of seven sensors used in this study. Two strain gauge sensors were mounted on the tool holder flat surfaces to detect the strain on the cutting and feed direction. Two piezoelectrics were mounted on the other two flat surfaces to detect vibrations in the cutting direction and feed direction. The tool holder was mounted to the force dynamometer that detects the change of force in three directions, cutting, feed and radial. Machining process was carried out by performing turning machining on the hardened carbon steel workpiece AISI H13 using TH1000 grade carbide tool (ISO grade H). The purpose of the turning machining process is to obtain four types of data namely strain, vibration, force (in time domain) and tool flank wear (VB) signals. The width of the flank wear is measured on each run until the flank wear reaches 0.3 mm. I-kaz and I-kaz multilevel signal features (SF) were extracted from the signals captured and then correlated to flank wear progression. A total of 14 SFs were assigned as the input parameter for the MIS prediction method. The MIS prediction method manage to predict correctly 55 out of 60 run of validation. This represent 92% prediction accuracy. With this accuracy, the system can be used as the tool to predict the flank wear condition during machining process