Partial Discharge Characteristics in Aerospace Applications Under High dv/dt Square-Wave Voltage Pulses

Abstract

The aviation industry faces a growing imperative to decrease greenhouse gas emissions and transition towards more electric aircraft (MEA) and all-electric aircraft (AEA). One key requirement for a successful transition is increasing the power density and efficiency of onboard power converters. Wide-bandgap (WBG) power switching devices, such as silicon carbide (SiC), present a promising solution due to their high voltage capability and rapid switching speed. However, these devices also pose a challenge as their fast switching speed can adversely affect insulation systems. This dissertation aims to investigate and offer a deeper understanding of, as well as solutions to, the challenges related to partial discharge (PD) in aircraft power wiring under high dv/dt voltage pulses generated by SiC devices.

The dissertation begins with an overview of the significant technical challenges related to PD that arise from using SiC device-based variable speed drives (VSD). It then discusses existing studies addressing these challenges and the failure of insulation systems due to PD. The literature review highlights disagreements in current research and the absence of comprehensive investigations into PD behavior caused by high dv/dt square-wave voltage pulses and their impact on the premature failure of insulation systems.

Subsequently, the dissertation delves into the challenges associated with the PD phenomenon in aviation wires, explaining the factors that influence PD behavior, accurate PD detection methods, and the extraction of meaningful features to quantify PD intensity. This is followed by an experimental study of PD-induced aging under various conditions to explore the effects of different variables on PD behavior and the failure mechanism of aviation wires. The experimental conditions are chosen to examine voltage rise time, voltage amplitude, ambient pressure, ambient temperature, and the dielectric material of aviation wires.

The experimental results reveal the complexity of PD behavior and the associated degradation process, highlighting the need for artificial intelligence (AI) and machine learning (ML) to interpret and predict insulation status. The dissertation then presents the implementation of ML, showcasing feature extraction, data preparation, model training, and the successful prediction of remaining service life. Finally, the dissertation concludes with a summary of key findings and recommendations for future research.