Haar-Wavelet-Based Statistical Scanning of Turn Fault in Vehicular Starter Motor

Abstract

Stator interturn fault is one of the most common faults that affect motor operation. Its failure will not only prevent the vehicle from starting but may also cause other adverse problems. In this letter, an automotive system has first been modeled in MATLAB/Simulink environment. The starter motor used in this system has been simulated to undergo a stepwise interturn fault. For every shorted starter armature coil, the current data were recorded and then wavelet analysis was performed on the gathered data. The wavelet-based parameters variation was determined for every degree of short on the starter stator armature coil. Based on learning, an algorithm has been proposed and then validated with real data to detect the variation in the wavelet parameters and to help determine normal operating conditions, prefault conditions, as well as conditions during fault of the automotive system to help in early detection of starter motor interturn fault. This has been done using a square-shaped Haar wavelet function based on the Hadamard-Haar transform. Such early detection can be very helpful for automotive systems, and if such faults are left unattended, it may result in complete machine failure due to excess heat generation due to the flow of high current in the shorted coils.