An Optimized Intelligent Technique for Bearing Fault Diagnosis using Motor Current Signal Analysis

Abstract

Bearing fault diagnosis is typically based on mechanical sensors, which are highly acceptable for large-size critical applications in the plant/industry. Condition monitoring (CM) of small size equipment, is highly challenging using mechanical sensors due to its additional cost and low diagnosis accuracy. CM using mechanical sensors is not feasible for several critical applications such as submersible pumps, extremely high/low system operating temperature, not accessible condition etc. Moreover, multiple sensors are required to diagnose the different types of faults in the equipment or subsystem, which further reduces mechanical sensor's reliability and affordability. In this paper, a novel approach using electrical signals is proposed for external bearing mechanical fault diagnosis without using mechanical sensors in a non-intrusive way. The performance analysis of diagnostic accuracy is demonstrated using the experimental dataset of real damage and artificial damage condition. The demonstrated result shows the high diagnosis capability of the proposed approach of 97.8% with EMD based selected features.