Identificação de deformidades em grãos de chips de resina PET com o uso de inteligência artificial

Abstract

This article addresses the use of machine learning to detect visual anomalies in PET (polyethylene terephthalate) resin chips, which may be caused by manufacturing, storage, or transportation issues. It has explanations about what machine learning is, PET resin and visual anomalies. It then discusses how machine learning is used in industry and what the most common techniques are, such as linear regression, logistic regression, decision trees, neural networks, and clustering algorithms. The article then presents a methodology for using machine learning to detect visual anomalies in PET resin chips, which involves creating a machine learning model trained on images of normal and abnormal chips with an accuracy of 97.53 %. The model can then be used to scan chip images in real time and alert operators for further investigation, improving the quality of the manufacturing process and reducing the number of defective chips produced.