기계학습을 기반으로 한 자외선 경화형 도장의 부착성 불량 위험수준 정량화

Abstract

Ultraviolet-curable coating is environmentally friendly and has the advantage of reducing the process time. The UV-curable coating is mainly applied to interior and exterior parts of automotive vehicles. In particular, adhesion is an important factor in the quality control of the automotive manufacturing process to improve the product reliability. Therefore, human being experts make an effort of testing the adhesion defect in the automotive manufacturing process. However, there is a disadvantage that the test result is subjective and costly because the engineer manually performs the adhesion test. In this study, the risk level of adhesion defect is quantified to predict the adhesion defect in a UV curable coating process. XGBoost is employed to predict the adhesion defect of a product and quantify the corresponding risk level in an automotive manufacturing process.