By Frank Sun, Dr. Bilal Akin, and Dr. Shuai Zhao
Our research aims to investigate incipient faults of components and create online fault diagnosis / failure prognosis tools to establish early warning system for power electronics systems. Continuously monitoring these systems is essential to prevent unexpected shutdowns and catastrophic failures that may result in fatal accidents or significant operation loss. In order to move power electronics technologies forward reliably, we investigate progressive degradations and parameter shifts in components as well as develops online degradation monitoring tools. These tools lay foundations of self-monitoring, smart-energy conversion systems that can recognize failure precursors at the earliest stage and prevent catastrophic failures.
In addition to our systematic physical failure inspections and failure analysis, we deploy machine learning / data driven prognostic techniques in PHM to unleashes great potentials for system predictive maintenance. In this webinar, PHM applications are categorized and the advantages and limitations of relevant machine learning algorithms are detailed. Subsequently, several examples of state-of-the-art research within intelligent condition monitoring, feature mining, detection, diagnosis, and prognosis will be presented. Finally, the presenters will share their perspectives on the existing challenges and future opportunities of PHM on power electronic systems.