Multivariate Time Series and Machine Learning Techniques for the multi-class Classification of Lathe’s Cutting Tools Wear Condition

Thiago E. Fernandes; Matheus A. M. Ferreira; Guilherme P. C. de Miranda; Marcos V. G. R. da Silva; Eduardo P. de Aguiar

doi:10.20906/CBA2022/3585

Thiago E. Fernandes Federal University de Juiz de Fora, Juiz de Fora, MG
Matheus A. M. Ferreira Federal University de Juiz de Fora, Juiz de Fora, MG
Guilherme P. C. de Miranda Federal University de Juiz de Fora, Juiz de Fora, MG
Marcos V. G. R. da Silva Federal University de Juiz de Fora, Juiz de Fora, MG
Eduardo P. de Aguiar Federal University de Juiz de Fora, Juiz de Fora, MG

DOI: https://doi.org/10.20906/CBA2022/3585

Keywords: Autonomous Learning, Empirical Data Analyses, Machine Learning, Machining Processes

Abstract

The majority of mechanical components went through a machining process during their manufacturing. Therefore, manufacturing processes with inadequate condition tools are likely to induce unexpected operational interruptions, accidents, product quality, and economic losses. Accordingly, the ability to classify fault imminences can result in cost reduction, along with productivity and safety increase. This paper aims to discuss an autonomous model based on the Self-Organised Direction Aware Data Partitioning Algorithm (SODA) and machine learning techniques, including time series Feature Extraction based on Scalable Hypothesis tests (TSFRESH), to solve this problem. The model proposed in this work can identify the patterns that distinguish the cutting tool’s flank wear in a multi-class scenario as adequate, intermediate, and inadequate conditions, achieving satisfactory performances in all cases and allowing to prevent fault occurrences.