Speaker: Changhong Zhao Faculty host: Professor Gil Zussman
Abstract: Cyber-physical systems, such as electricity, water, and transportation networks actuated by computing algorithms on data networks, compose the critical infrastructure underpinning our nation’s economy, security, and health. Current frameworks to operate the cyber-physical infrastructure are often decoupled across different timescales, objectives, and sectors and are often centralized. They therefore lack the responsiveness and flexibility to meet future needs for security, resilience, and efficiency, under emerging challenges including larger system scales, increased variations from renewables, sever cyber-physical threats, and complex interdependencies across sectors. To overcome these challenges, my research aims to develop integrated, distributed, and computationally efficient control methods for interconnected cyber-physical infrastructure. In this talk, I will present two research results on power systems, for which the phenomena, problems, and solutions are representative in the broader cyber-physical field.
The first project is integrated design of network optimization and feedback control with application in power system frequency control. We developed a distributed control scheme that can stabilize a power network at an optimal equilibrium satisfying frequency and power flow security requirements at minimal cost. Compared to the traditional centralized approach that separates optimization and control, the proposed scheme is more scalable to large networks, more resilient to cyber-physical threats, more adaptive to large and fast variations, and economically more efficient. We proposed a novel systematic approach to designing such control. The key idea of this approach is formalizing control goals as an appropriate network optimization problem and deriving closed-loop dynamics with distributed feedback controllers as a real-time primal-dual algorithm to solve the target optimization and its Langrangian dual.
The second project is convex relaxation of optimal power flow in multiphase unbalanced networks with both wye and delta connections. For that, a semidefinite reformulation/relaxation combined with a linear approximation of delta connection is developed and is shown to be exact and accurate by numerical results. The convex relaxation approach can also be applied to an integrated optimization of power and water networks, which demonstrates significant cost reduction compared to the traditional decoupled operations of the two networks.
Bio: Changhong Zhao is a Lead Researcher at the National Renewable Energy Laboratory. He received the B.Eng. degree in Automation from Tsinghua University, Beijing, China, in 2010, and the PhD degree in Electrical Engineering from California Institute of Technology in 2016. He was a recipient of the Caltech Demetriades PhD Thesis Prize and the Caltech Wilts PhD Thesis Prize. His research focuses on distributed control and optimization of critical cyber-physical infrastructure, especially power and energy systems.