Detection of Damaged Inserts of Cutting Tools Using Deep Learning

Abstract

Industrial manufacturing has become increasingly mechanized in recent years. The manufacturing productivity and costs of products are significantly impacted by machine cutting tools, which are a crucial component of industrial production. Tool breakage frequently happens suddenly and without warning in a practical manufacturing process, leading to an unusually unbalanced ratio of tool breakage samples to normal samples. Considering the need of current scenario of development of smart system in production units, the authors have proposed a deep learning-based model for prediction of damaged inserts which can give accurate results as compared to traditional techniques and manual inspection methods. The real time data of damaged tool and undamaged tools were collected, and the model training was done to predict defective inserts with high accuracy.