Detection of Broken Rotor Bar Fault in an Induction Motor Employing Motor Current Signature Analysis

In the present industrial scenario, an induction motor plays an important part for any industry. Any faults in induction motor may lead to expensive and catastrophic for a plant, which may directly influence the reliability of the whole system. Also, rotor faults constitute an important portion of all electrical machine faults and can be a key reason for downtime of the system. Therefore, rotor fault diagnosis of an induction at an early stage can avoid the costly breakdown in an industrial plant. The present work proposed a novel detection algorithm for broken rotor bar in an induction motor using motor current signature analysis (MCSA) which is able to detect fault at an early stage. Real-time motor current from stator winding of an induction motor was acquired and developed algorithms were applied for the detection of broken rotor bar fault. Previously, other sensors such as vibration sensors were also used for this purpose but, those are invasive in nature with high cost however, current sensors are non-invasive and low cost which allow the current sensors-based techniques easier to implement. In this work, statistical method such as multiscale entropy (MSE) algorithm is used as a novel technique which is extracted from the real-time motor current for the detection of broken rotor bar fault. Later, support vector machine (SVM) has been successfully implemented to diagnose the fault at early stage with minimum accuracy of 90.1%.