To improve energy and environmental sustainability, the power grid has been increasing the penetration of distributed energy resources (DER), such as solar panels, wind turbines, electric vehicles, responsive demands, etc. Combining two-way power flow with two-way communication flow, these modifications can cut peak energy use, drive energy conservation, replace traditional electricity reserves, enable renewable resources, and reduce greenhouse gas emissions. However, deploying DER in a widespread, cost-effective, and reliable manner requires complex integration with the existing electricity grid due to challenges including i) a large number of DER, ii) the volatility and uncertainty in the physical systems, renewable generation, and user consumption, iii) the partial controllability of DER due to human interaction, and iv) the still-undeveloped sensing, communication, and computation resources required to support the future grid. To address these challenges, this project has harnessed the structure of the power systems to develop a suite of distributed coordination algorithms and mechanisms to control and incentivize the DER with provable performance so as to facilitate a smoother transition for the grid into the next age of a smarter grid.

Specifically, this project has designed automated distributed algorithms to optimize the DER performance at slow-time scales and control the DER to ensure the energy balance at fast-time scales, and designs incentive schemes such as pricing, rewarding, payoff, and trading rules to promote human participants to take desired actions. Distributed automated algorithms tackle the challenges brought by the power system physics (e.g. physical laws such as Kirchhoff's law and system dynamics such as swing dynamics), limited communication, and uncertain DER generation and consumption.  To handle the challenges of the uncertainty in both the physical systems and human, this project has also developed a variety of online learning, machine learning, and reinforcement learning methods. The unique intellectual merit of this project is the cognizant design of coupled physical systems, cyber (sensing, communication, computation, and market) networks, and the decision-making algorithms. The developed methods maximize the use of physics in order to lower sensing, communication, and computation cost and to limit agents’ market power as the local physical variables carry a large amount of information about the global network conditions. This project has integrated mathematics, computation, engineering, and economics to jointly design the distributed architecture, algorithms, and incentive schemes to ensure efficient, adaptive, and reliable DER coordination for future power networks. The developed algorithms have been tested on real test beds by national labs (for microgrids), architectures (for green buildings), and computer engineers (for data centers). Some of them already have been implemented by industry on real-world field studies (residential DR).

For broader impacts, all the research results have been disseminated through conference and journal publications,  to accelerate the transition of the results to industry, the PI and her former postdocs have launched a start-up company, Singularity Energy Inc,  providing AI-empowered digital solutions on decarbonization of power grids. Moreover, the developed results are applicable to a broad class of sociotechnical systems, such as green buildings, air/urban transportation systems, water/gas distribution systems. For advising and education, the PI has been actively advising a group of undergraduate students, graduate students, and postdocs, most of whom are pursuing academia careers after graduating from the PI's group. In particular, 2 former graduate students and 2 former postdocs who were advised under this project are tenure-track assistant professors. Lastly, the PI has been actively participated in and organized many activities in promoting STEM in K-12 education. As one example among all the activities, the PI has volunteerly advised a group of elementary school girls on coding and lego robots through weekly hands-on activities. 

Last Modified: 03/06/2023
Modified by: Na Li