Bearing Fault Diagnosis Using Motor Current Signature Analysis and the Artificial Neural Network

Abstract

Bearings are critical components in rotating machinery. The need for easy and effective bearings fault diagnosis techniques has led to developing different monitoring approaches. In this research, however, a fault diagnosis system for bearings is developed based on the motor current signature analysis (MCSA) technique. Firstly, a test rig was built, and then different bearing faults were simulated and investigated in the test rig. Three current sensors, type SCT013, were interfaced to Arduino MEGA 2560 microcontroller and utilized together for data acquisition. The time-domain signals analysis technique was utilized to extract some characteristic features that are related to the simulated faults. It was noticed that the simulated bearing faults have led to generating vibrations in the induction motors, which in turn cause a change in its magnetic field. For classification (identification) of the extracted features, the artificial neural network (ANN) was employed. An ANN model was developed using the Matlab ANN toolbox to detect the simulated faults and give an indication about the machine health state. The obtained features from the captured motor current signals were utilized for training the ANN model. The results showed the effectiveness of using MCSA based on the time-domain signal analysis in combination with ANN in diagnosis different bearings faults.