Wireless Hybrid Vehicle Three-Phase Motor Diagnosis Using Z-Freq Due to Unbalance Fault

Abstract

Online diagnostics of three phase motor rotor faults of hybrid vehicle can be identified using a method called machine learning. Unfortunately, there is still a constraint in achieving a high success rate because a huge volume of training data is required. These faults were represented on its frequency content throughout the Fast Fourier Transform (FFT) algorithm to observe data acquired from multi-signal sensors. At that point, these failure-induced faults studies were improved using an enhanced statistical frequency-based analysis named Z-freq to optimize the study. This analysis is an investigation of the frequency domain of data acquired from the turbine blade after it runs under a specific condition. During the experiment, the faults were simulated by equipment with all those four conditions including normal mode. The failure induced by fault signals from static, coupled and dynamic were measured using high sensitivity, space-saving and a durable piezo-based sensor called a wireless accelerometer. The obtained result and analysis showed a significant pattern in the coefficient value and distribution of Z-freq data scattered for all flaws. Finally, the simulation and experimental output were verified and validated in a series of performance metrics tests using accuracy, sensitivity, and specificity for prediction purposes. This outcome has a great prospect to diagnose and monitor hybrid electric motor wirelessly.