In fabrication processes are cutting tools used to shape raw material into (end) products. Especially in metal cutting can tools wear out within minutes, mostly depending on tool material, workpiece material and the cutting speed. Difficult to machine workpieces, such as stainless steel, which will be used in this report, can increase the complexity to model and predict wear, because tool life will be a lot shorter compared to easy to cut workpieces. This is due to low thermal conductivity, high work hardening rate and adhesive/sticky characteristics of the material which produce higher cutting forces and temperatures which result into increased tool wear rates [1].

Cutting using worn tools will lead to reduced component quality, whereas preventive tool replacement will increase the production costs. Optimizing the tool replacement strategy is crucial for an efficient and sustainable fabrication process. 

Many companies use eighter a fixed time period to replace the tools, or an experienced operator determines when the tool will be replaced based on surface roughness, cutting force, motor force, acoustic emissions, vibrations or tool inspection [2]. Both of these methods are not optimal and depending on the operators, will the replacement strategy be inconsistent.

In previous works has already been defined that the tool wear can be estimated by performing audio analysis [3] [4]. These studies used a generic audio measurement system to record the sound waves. Data from the cutting processes have been analyzed in a preliminary investigation project to see if a machine learning algorithm would be able to estimate the tool state. During that investigation was found that the tool wear could be defined as “usable” or “replace”. The results of this investigation looked promising but the process required a lot of manual tasks which consisted of recording, saving, converting and analyzing the audio. Due to the fact that the tool state could only be defined after the cutting, rose the idea of real time automated tool wear analysis. That idea leaded to the study in this report.

The aim of this thesis is to create an analysis system that will determine when the tool should be replaced. This system will be completely automated, and will estimate the tool state during the cutting process.

In this study will a system be built to analyze the tool wear of an automated Computer Numerical Control (CNC) lathe cutting grade EN 1.4307 (AISI / ASTM 304L) stainless steel.