Analysis of Cutting Tool Condition Monitoring System for Turning Operation

Abstract

Cutting tool wear monitoring in machining operations has been an active area of research for nearly last two decades. Cutting tool wear plays an important role in deciding economic strategies, product quality, tooling cost, tool-changing cost, rejection of products and productivity. Metal cutting processes are in general non-linear and stochastic in nature. It is therefore difficult to represent them as a mathematical model and they usually require simplifying assumptions. As a result, such models are not capable of representing real metal cutting process. For an automated industry, all the machining input parameters (cutting speed, feed rate, depth of cut) are controllable except cutting tool condition. Major problem in the machining process is cutting tool wear prediction. In this research work an attempt has been made to develop neural network models using sensors’ signals to predict the health of a cutting tool. Properties of signals from the sensors depend on many factors such as machining conditions (cutting conditions), workpiece material, and cutting tool geometry. Apart from the complexity of the process, signals from the sensors are disturbed for many reasons: outbreak at cutting edges, chatter (i.e. self-exited vibrations), sensor non-linearity, noise of digitizers, crosstalk effects between sensor’s channels, etc.