MECHATRONIC SYSTEM BASED ON NEURAL NETWORKS FOR THE DETECTION AND CLASSIFICATION OF DEFECTS IN CERAMIC PARTS IN FACTORIES

Abstract

The proposed system automates the detection of defects such as cracks, stains, and irregular edges using image processing algorithms and machine learning techniques. A prototype was developed that includes a camera, a lighting system, and software that processes the captured images. The results show a significant improvement in defect detection, reducing waste and production downtime. This system not only optimizes the inspection process but also provides a low-cost and adaptable solution for the ceramics industry, highlighting the importance of technological innovation in modern manufacturing. This real-time analysis allows for immediate decisions on the production line, such as the removal of defective parts.The project offers highly favorable results, as it will allow, on the one hand, the classification and disposal of defective ceramics and, on the other, the storage of all related information, both accounting and descriptive, including the quantity of discarded ceramics and the defects they presented. This aims to facilitate monitoring by those in charge and optimize decision-making at the plant, identifying processes that require greater attention.