Diagnosing simultaneous bearing and misalignment faults in an induction motor using sensor fusion

A monitoring system for induction motors (IMs) is essential for most industrial plants. Bearing faults and shaft misalignment are common mechanical defects in induction motors. Since one fault could cause another fault in the system, multiple faults can occur simultaneously and change the vibration (electrical) behaviour of the induction motors from that of a single fault condition. This paper aims to identify two common faults (shaft misalignment and defective bearing) simultaneously in IMs using data fusion of vibration and current measurements. Sensor fusion of accelerometer and Hall-effect sensor signals is used to combine the vibration and current signals. The proposed method is applied via a laboratory test-rig based on data fusion to detect multiple defects simultaneously in induction motors. Then, by extracting the important features using a principal component analysis (PCA) algorithm, the K-nearest neighbours (KNN) classification algorithm is used to detect defects and make decisions. The results show that the fusion of both current and vibration signal analyses significantly improves the efficiency and reliability of multiple fault detection. Also, bispectrum analysis of the current signal is highly sensitive to misalignment and can be an effective method for detecting such faults.