About the project
Objective
This project aims to develop a theoretically grounded, data-driven congestion control framework that combines output regulation, online feedback optimization, and reinforcement learning to achieve autonomous and optimal congestion mitigation in power grids, with guaranteed robustness and adaptability to renewable generation uncertainties, without relying on explicit grid models.
Background
The electrical power grid is undergoing a profound transformation driven by digitalization, market deregulation, and the rapid integration of renewable and flexible energy resources. The EU Green Deal and Sweden Climate Policy Framework set ambitious targets for a sustainable and resilient energy system, while addressing energy poverty. Yet, unlike traditional centralized plants, renewables are dispersed and weather-dependent, pushing the grid close to its operational limits. When congestion occurs, overloaded lines threaten stability and clean energy is curtailed. Industry manual and semi-automated congestion management mechanisms cannot cope with this growing complexity. The recent large-scale blackout across the Iberian Peninsula has revealed the fragility of existing operational paradigms and the urgent need for the integration of more efficient autonomous, data-driven control. This raises a fundamental question: How can we design autonomous, data-driven congestion control algorithms that harness the increased digitalization of the power grid to ensure efficient and reliable operation despite the volatility of renewable generation? To address this question, this project aims to develop a robust online reinforcement learning framework, grounded in output regulation theory and online feedback optimization, that learns optimal congestion mitigation strategies directly from real-time power grid data without relying on an explicit grid model.
About the Digital Futures Postdoc Fellow
Liquan Lin received his PhD from The Chinese University of Hong Kong, where his research focused on developing data-driven control methods for uncertain linear systems, with particular emphasis on the output regulation problem. His current research interests lie at the intersection of output regulation theory and feedback optimization, with a focus on data-driven methods and their applications to power systems.
Main supervisor
Giuseppe Belgioiso
Co-supervisor
Qianwen Xu
