CAREER: Advancing Autonomy of Power Grids in 3D -- Decision Speed, Disruption Resilience, and Distributing Societal Benefits

This NSF CAREER project aims to advance the autonomy of power grids by developing the fundamental theory for enhancing decision speed, resilience, and societal and sustainability awareness of distributed grid management models and algorithms. The fusion of ubiquitous autonomy and connectivity, along with the widespread adoption of distributed energy resources, is transforming our legacy grid into a multi-agent and distributed infrastructure.

The project brings transformative change to power grid management by developing innovative strategies to orchestrate the autonomy of cyber-enabled components (agents) to enhance the grid's efficiency, resilience, and sustainability. This is achieved by addressing three critical research questions: how to leverage agents’ autonomy for faster decision-making, enhance the resilience of managing a multi-agent power grid, and integrate societal awareness and carbon footprint considerations into distributed power management models.

The intellectual merits of the project include the development of novel machine learning (ML)-assisted optimizers to rapidly solve complex decision-making problems for individual power grid agents, the use of Artificial Intelligence (AI) and generative modeling to improve the resilience of collective power grid agents, and the design and integration of societal benefits and sustainability metrics into the multi-agent decision-making paradigms. The broader impacts of the project include enhancing power grid performance and resilience, educating the public through various media (print media and broadcast news), and offering educational and research opportunities for students, government, and industry professionals.

The goal of this project is to enhance the efficiency of the power grid whose characteristics are increasingly multi-agent and distributed in nature. The efficient operation of future electric infrastructure depends on leveraging agent-level autonomy to enhance grid management functionalities in a timely, resilient, socially aware, and sustainable manner.

To achieve this, the project proposes neural approximators for mapping optimization inputs directly to high-quality, feasible outputs, thereby eliminating agent-level iterations and significantly speeding up agents’ computations. In addition, this project seeks to enhance the resilience of a collection of agents by using AI to regulate inter-agent communication parameters, using generative models to address data gaps, and dynamically adjusting information dissemination rates during/post disruptions.

Finally, the project plans to incorporate societal considerations into distributed power dispatch models by first studying how design parameters, such as computation granularity and strategies for managing struggling agents, affect the distribution of societal benefits and the carbon footprint of distributed computations. These findings subsequently guide the development of socially aware and sustainable models for distributed grid management.

The expected outcome of this research is a set of novel analyses accompanied by algorithms, tools, and techniques that harness the autonomy of power system agents to enhance the efficiency and resilience of the nation’s power grids.

This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.