I will work on large-scale integration of renewable energy resources such as rooftop solar PV: specifically, I will develop state-of-the-art tools at the intersection of machine learning and power system control to safely and efficiently integrate distributed energy resources.
A motivating application for me is the integration of solar PV into the distribution system. They can both create large and rapid voltage fluctuations, as well as correcting them using inverter interfaces. Optimizing the actions of inverters is complex and computationally very challenging. Machine learning can be very helpful in designing the controllers but due to the uncertainty and volatility of the system, existing methods do not guarantee stability and safety. I will develop efficient voltage controllers that are provably safe. I will use Lyapunov theory to characterize the space of stable controllers. This constrains the search space of deep learning and reinforcement learning to guarantee safety.
Advisor: Baosen Zhang – Electrical & Computer Engineering