About

The modernization of power systems for clean energy by integrating multiple renewable resources is changing the dynamics of power grids at a fundamental level. There is a dire need to understand new phenomena and possible failure mechanisms to unlock the design of countermeasures so that operators can make electric grids more resilient. But the required degree of understanding must keep up with the pace of new technologies in generation and storage, sensing and communications, optimization and control, power electronics, machine learning, and data science. This NSF project aims to develop a unified framework towards this goal, starting from sensors to algorithms to real-time control. The project will bring transformative change by leveraging fundamental developments in control, power electronics, and machine learning, and fusing them with trusted power system models, significantly enhancing the ability to predict and control grid dynamics with a high share of renewable energy resources. Results will be verified by building a digital twin of a large-scale transmission grid. The intellectual merits of the project include a balanced solution between models of renewable-integrated power systems developed from first principles and those identified from data, and the convergence of advanced methods under development within otherwise disconnected research communities. The broader impacts of the project include addressing pressing research questions whose solution will enable the building blocks of a cleaner power grid. The project will also engage underrepresented groups in STEM.

A central problem hampering the pace at which one can integrate renewable energy sources into electric power grids is the insufficient understanding, at a systems level, of the dynamic interplay between existing assets and inverter-based resources (IBR) deployed at scale on a transmission grid of substantial size. This project will address this challenge by creating a unified modeling environment for bulk transmission grids that integrates data-driven yet analytical IBR models. The resulting framework lends itself seamlessly to a state-space form familiar to those working with dynamical systems. Thus, the proposed framework is inclusive beyond traditional disciplines in power systems modeling. The approach will be to leverage this inclusiveness by absorbing into a digital twin of a transmission grid the latest developments in tangential areas driving innovations in power systems.

Research areas

Power Electronics

Dynamics

Koopman operator.

Power Electronics

Control

Reinforcement learning.

Power Electronics

State Estimation

Robust dynamic state estimation.

Power Electronics

Sensors

Battery state of charge.

Power Electronics

Power Electronics

Data-driven impedance model.

Power Electronics

Machine Learning

Deep operator networks.

Team

Marcos Netto

Dynamics

Na Li

Control

Junbo Zhao

State Estimation

Philip Pong

Sensors

Minjie Chen

Power Electronics

Guang Lin

Machine Learning

Advisory Board

Vahan Gevorgian

NREL

Chris King

Siemens

Chad Watson

PSEG

Aditya Ashok

OPAL-RT