Efficient Machine-Learning Model for Bearing Fault Identification Using the CWRU Dataset

The deterioration of rolling element bearings significantly impacts the performance and longevity of a system. Regarding mechanical power transmission systems and other electromechanical drive systems, rolling element bearings are an essential component of many different types of industrial equipment. In a rolling element bearing, a few issues can happen in the outer race, the inner race, the cage, or a rolling element. Several studies show that faulty bearings cause 40% to 70% of electromagnetic drive system breakdowns. In this work, machine-learning (ML) techniques are used to diagnose and identify the rolling element bearing. The Case Western Reserve University (CWRU) website is where the vibration dataset for the rolling element bearing was obtained. Preprocessing is done on the datasets to manage missing values, eliminate noise and outliers, and normalize the data. The traits of the information are then taken out to make useful data. This dataset is used to choose, train, and test different machine-learning models. The result shows that the proposed method of keeping the bearings in good shape and ensuring the induction motor runs safely works.