Abstract

Electrified vehicles commonly use traction machines powered by voltage source inverters (VSI) for efficient speed and torque control. However, short circuit faults and insulation failures remain prevalent, accounting for approximately 30% of motor failures. Given the uncertainties surrounding insulation degradation, detecting degradations of insulation in an early stage can help prevent major failures. Therefore, this Ph.D. research focuses on online monitoring of electrical machine’s winding insulation degradation.
A comprehensive review of literature revealed certain research gaps. The first one is on selection of the most effective insulation degradation indicator for online condition monitoring without increasing motor drive costs. To address this challenge, this research uses existing signals in EV motor drives, such as line current measurements. However, there is limited information on how insulation degradation can impact the line currents in the existing literature. Therefore, this Ph.D. work address this knowledge gap through conducting investigations of insulation indicators. It is found that the antiresonance oscillations in line current can serve as indicators for insulation degradation, which was not reported in the existing literature.
Existing literature on condition monitoring methods also presents notable limitations. Firstly, these techniques can not determine the degradation of groundwall (GW) or turn-to-turn (TT) insulations simultaneously. There is a need for a new approach for simultaneous condition monitoring of TT and GW insulations. This is crucial because different types of insulation are exposed to different temperatures, leading to a varied degradation rate. Additionally, current methods overlooked the variability of noise in measured signals, which can fluctuate due to various factors in real-world applications like EVs. This variability necessitates a condition monitoring approach that can handle noise while accurately determining insulation health. Moreover, existing methods rely on predefined thresholds and manual analysis, requiring expert interpretation, which limits their applicability across different machines and conditions. Hence, this Ph.D. work proposes novel methodologies to address these limitations. A technique for simultaneous monitoring of TT and GW insulation conditions has been proposed. To address the limitation posed by noise variability and the reliance on manual analyses, a novel data-driven methodology for robust insulation condition monitoring has been proposed.