Rolling element bearing fault diagnosis using machine learning techniques: A review

Smart factories are enabled with heavily digitalised and networking systems through machine learning and Artificial intelligence techniques as a heart of smart factory technology. The machine learning and also deep learning algorithms provides benefits in speech recognition, object detection, image processing, fault detection as well as in medical science. The Machine learning models mainly widely used in machinery defect detection and diagnostic (FDD) systems in recent years. The automated feature learning procedure of deep architecture has a lot of promise for solving problems with traditional fault detection and diagnosis systems (TFDDS). The purpose of this review is to study and critically analyse different machine learning fault diagnosis approaches used in rolling element bearings. It may also aid in a better understanding of the processes and procedures used in all types of defect detection machine learning techniques, providing fresh insights for future study that may lead to improved rolling element bearing performance. Early and precise diagnosis of the problem is critical, since it can help to reduce wear and tear of the equipments. Finding a method that can deliver a high level of precision and accuracy in early diagnosis will result in a significant improvement in machine performance.