A review on bearing fault diagnosis methods with emphasis on machine learning and deep learning applications

To improve plant efficiency, reduce unnecessary costs, and enhance safety, industries must adopt effective predictive strategies for supply management optimization. Since more than 90 % of machines rely on rolling element bearings, their failure often results in significant machinery breakdowns. In fact, a considerable share of failures in manufacturing systems can be attributed to bearing defects. This review focuses on examining and evaluating various diagnostic methods for rolling element bearings, comprising conventional approaches, machine learning techniques, and advanced deep learning methods. By delivering a comprehensive understanding of these detection strategies, the study aims to support better implementation practices and offer fresh insights for future research directed toward enhancing the reliability and performance of rolling element bearings.