Data-Driven Dynamic State-Estimation for Modern Power Systems

Lehigh University, Bethlehem PA

Investigators

Abstract

This NSF project aims to develop data-driven algorithms for dynamic state-estimation of modern power systems with uncertainties of distributed energy resources. The project will bring transformative change to the state-of-the-art dynamic state-estimators that heavily rely on accurate physics-based models and are challenged by model uncertainties and disturbances. A combination of machine learning and signal processing techniques will be used to convert available measurements into accurate dynamic models. The intellectual merits of the project include generating new knowledge on monitoring and situational awareness in modern power systems as well as developing a novel data-driven paradigm to improve smart grid resilience through dynamic state estimation. The broader impacts of the project include addressing grid outages and resolving cascading failures by better tracking power system asset dynamics in real-time. A number of educational and outreach activities are also used to address the lack of students in electrical engineering, such as summer research activities for students through STEM Summer Institute (STEM-SI), custom-designed senior projects, and open-source materials available to the scientific community. It is important to develop efficient dynamic estimation techniques to better monitor and control inverter-dominated grids, but there is a technical gap in modeling distributed energy resources (DERs) that is not fully understood. To address this challenge, this project will develop advanced data-driven approaches leveraging statistical machine learning theory and available measurements to identify nonlinear mathematical models of inverter-based DERs. Using the developed models, uncertainty-aware decentralized data-driven dynamic state estimation of DER states will be designed without the need for complex physics-based models or simulations. In addition to providing accurate and state-aware dynamic models for various types of DERs (such as storage, solar panels and wind generators), the research results from the proposed framework will reduce the current complexity of implementing dynamic state estimation. 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.