Detection of tool breakage during milling process through acoustic emission

Abstract

Catastrophic tool failure (CTF) in milling process can cause damage to the product’s machined surface and the machine tools, leading to huge financial losses. It is therefore critical to detect CTF in advance and promptly respond to it. Because of the safety and quality requirements imposed in practice, there are far fewer failure samples than normal samples, and this disequilibrium makes it difficult to detect failures. The aim of this study is to develop a new, easy, and practical automatic system for tool breakage detection using the acoustic emission (AE) technique. Components of AE raw data are analysed to locate the moments of tool breakages and to screen the corresponding AE feature samples. A support vector machine-based cost-sensitive breakage detection model is established and optimized. The proposed model is applied and validated by experiments conducted on a factory’s milling machine. The model achieves an accuracy of 91.18% in the detection of breakages. The results show the practicability and validity of the proposed method.