Machine learning-based non-destructive method for identifying defect causes in OLED displays to enhance productivity

Eliminating defects in manufacturing is imperative for enhancing productivity and generating profits for companies. Numerous research studies on defect detection have been conducted to ensure product quality. However, improved productivity cannot be expected unless the cause of the defect is studied and actions are taken to eliminate it. In this study, we propose a method to identify the root cause of detected defects in display products. Traditionally, to identify the root cause of defects, products were destroyed and detailed analyses were conducted using high-performance equipment. However, this method reduces the commercial value of the products and decreases the revenue of the company. To address this issue, we propose an anchor process that utilizes data generated from the products. The proposed method offers the advantage of eliminating the need for product destruction. Furthermore, it provides an effective solution for managing noise that may occur during data collection and labeling, thereby enabling the practical implementation of machine learning theories in industrial applications. When the proposed method was used to predict the cause of the defect, the results were found to be consistent with the actual cause, thereby confirming the reliability of the method.