Join us on Zoom: https://ucsc.zoom.us/j/91506787166?pwd=TWFNSEplcUlCenRTd3phdXUrTDNJQT09 / Passcode: 577399
Description: Electric power grids are undergoing dramatic transformations that have brought many challenges regarding the situational awareness of the system. This research aims at developing efficient and robust strategies to deal with those challenges, in particular for three emerging problems.
First, we develop a data selection framework for classification and regression under real-time changing environments by performing theoretical investigation and empirical analysis in tandem. Our focus is on electricity price trend forecasting, and joint events and cyber-attack discrimination.
Second, we propose a reservoir-computing aided quasi-steady state estimation framework, which is resilient to topology changes and low-quality measurement coverage areas. We hypothesize that reservoir computing can capture the nonlinear dynamics of the power flow mappings and the deviations due to the intermittency of distributed energy resources (DERs).
Finally, we survey the challenge for microgrids faults detection via a model-based nonlinear observer to identify abnormal signatures present in the system. The growing penetration of DERs and power electronics based subsystems reduces the electrical inertia in microgrids, which makes fault detection more difficult. We will perform extensive experiments for the three problems by using IEEE benchmark test cases to demonstrate that our approaches are reliable and scalable.