Statistical Feature Based Identification of Rotor Fault Indicators for Three Phase Induction Motor

This paper presents a simple yet efficient method for diagnosing some major rotor faults using statistical tools. Proposed method has been developed by employing some simple statistical features through Motor Current Signature Analysis (MCSA) of induction motor operating under different conditions. Conventional technique such as Extended Park’s Vector Approach (EPVA) has also been involved in the process to transform 3-phase line currents into the form of Park’s Vector Modulus (PVM). Presence of oscillation in the Park’s Vector Modulus has been detected and analyzed with the help of statistical tools to get conclusive results regarding presence of rotor fault in three phase Induction motor. The artificial intelligence [fuzzy logic] using statistical tools or features like Kurtosis, Skewness, Cross correlation and standard deviation has also been incorporated for the classification of faulty and healthy conditions from the extracted feature set. This method gives true result in nearly all the cases [>90%] that can easily be observed in confusion matrix.