A Study of Traditional and Deep Learning Techniques for Fault Diagnosis of Bearings: A Survey
Abstract
In this review paper, we have summarized the overview of the available literature on fault
diagnosis of bearing with traditional and deep learning techniques. Traditional machine
learning (ML) techniques like support vector machine (SVM), artificial neural network (ANN)
and principal component analysis (PCA) has been used for the diagnosis of bearing fault in the
past. However, recently Deep Learning techniques have been developed as an intelligent
machine health condition monitoring. In the current paper, a brief review of traditional machine
learning methods are conducted and then moved on with the exploration of deep learning
techniques. We have studied that deep learning techniques performs feature extraction process
automatically unlike the conventional methods which requires a subject expertise to solve the
issue. Also many advantages of using deep learning technique has been summarized here along
with the discussion and suggestions on future work