Melhorias ao processo de pintura automotiva por meio de técnicas de machine learning

Abstract

Automotive paint automation is a complex process, consisting of multiple steps that apply sensitive layers of paint to the vehicle body in an orderly manner. Each layer has a particular purpose, which can be protection or final visual aspects. Due to the interdependence between paint layers, errors in early stages compromise subsequent ones. Thus, when the vehicle is submitted to the quality inspection, it may be disapproved for not meeting the required quality standards, being sent for rework, generating costs, in the automotive domain, of substantial impact. This article presents a machine learning approach to identify possible cost reduction opportunities in the automotive painting process, which could result in poor formation of the final product. The approach is implemented using clustering techniques, and applied to a real database of a Renault painting process.