Comparative Analysis of Machine Learning Techniques for Bearing Fault Classification in Rotating Machinery

Abstract

This paper provides a comprehensive analysis of techniques used for bearing fault classification, which is essential for detecting anomalous conditions in rotating machinery. The focus is on identifying and categorizing various types of bearing faults to monitor equipment performance and prevent repairable motor breakdowns. The authors use experimental data to identify bearing faults and extract significant features from the dataset, and then apply Principal Component Analysis (PCA) and Curvilinear Component Analysis (CCA) techniques for exploratory analysis. The study compares the classification accuracy of various machine learning models, including support vector machines, k-nearest neighbors, ensemble models, and neural network models such as Multilayer feedforward neural network (ANN) and Convolutional neural network (CNN). The results of this study provides valuable insights for future research in bearing fault classification since it is the most important component in rotating machines..