Prediction of single point cutting tool wear with cutting force signals using artificial neural networks

Abstract

Machining is used to create a variety of features like holes, slots, pockets, flat surfaces, and even complex surface contours on a given components like metals, plastics, composites, and wood. Hence, it is often considered as the most common and versatile of all manufacturing processes. The state of a cutting tool is an important factor in any metal cutting process. During the turning process, the tool enters and exists for cutting operation each time, so it is subjected to mechanical as well as thermal loading. The tool material gradually starts to wear or fractured and hence it is necessary to change the tool. Hence, the continuous monitoring of the cutting tool is necessary in order to replace the tool at right time without affecting the product quality and production rate. In this work, an effective low cost method of measuring the cutting force is designed and developed using the strain gauges. The work piece material used for turning is EN8 medium carbon steel and the tool material is made of coated carbide. The strain gauge with full bridge type II configuration is designed and developed to measure the bending strain. It consists of four active strain gauge elements, two strain gauges are mounted in the direction of bending strain and other two are mounted on the other side for measuring the principal axis of strain. The data acquisition system is used to capture the signals and is converted into equivalent cutting force values. An empirical model is developed for the responses cutting force using Response Surface Methodology (RSM). Tool wear, surface roughness and cutting force is considered as output responses. newline