The U.S. electric grid systems are increasingly incorporating smart grid sensors, energy storage, microgrids, and distributed energy resources such as solar photovoltaic panels. This research project will address the technical challenges in optimization under uncertainty for modeling the design and control of cyber-physical systems.
The design and control decisions of complex systems have been successfully modeled as optimization problems that often involve uncertain model parameters. Traditional approaches such as stochastic optimization and optimal control assume that the underlying distribution of uncertain parameters is known. Consequently, the resulting decisions are sensitive to changes in the distribution and inconsistency in time. This research project aims to develop a scalable data-driven optimization framework that enables statistically robust simulation of long-term operations of complex systems with uncertain model parameters, a capability that is central for the reliable design and planning of such complex systems.
High-performance algorithms will be developed for solving such problems. Adaptive decomposition methods for the algorithms will exploit their mathematical and statistical structure for efficient use on emerging computer architectures. The proposed research will improve reliability and resilience in the design and operation of the U.S. electric grid, edge computing networks, and other cyber-physical systems.