Comparative Study of Machine Learning and Ensemble Learning Approach on Tool Wear Classification

This study investigates the application of machine learning algorithms for predicting tool wear in machining operations, aiming to enhance production efficiency and reduce costs associated with tool maintenance. We implemented five distinct algorithms: K-Nearest Neighbors (KNN), Decision Trees, Random Forests, LightGBM, and XGBoost. The results reveal that these models can accurately classify tool conditions as "worn" or "unworn," with LightGBM and XGBoost showing solid performance. Notably, an ensemble approach using a soft voting classifier combining KNN, Random Forest, and LightGBM achieved an accuracy of 0.9968 and a ROC AUC of 0.9998. This research underscores the potential of machine learning to transform traditional tool management practices, enabling proactive maintenance strategies that can significantly improve machining efficiency and product quality. Future work may explore integrating real-time data for further enhancements in predictive accuracy.