Digital twin-based predictive control framework for DC microgrids

Abstract

Modern power electronic systems, essential for renewable energy and electric transportation, demand increasingly high levels of performance, reliability, and autonomy. However, achieving these goals is hampered by challenges in advanced control, modeling technologies and real-time health monitoring. This thesis addresses these challenges by developing and validating novel frameworks that integrate Digital Twin (DT) technology, predictive control, and Physics-Informed Machine Learning (PIML), with a specific focus on the Dual-Active-Bridge (DAB) converter.

The core contributions including three parts. First, a DT-based Predictive Control (DTPC) framework is proposed, which leverages a high-fidelity virtual model to achieve superior dynamic response and robust voltage regulation compared to conventional methods. The framework's efficacy is rigorously demonstrated through simulation, Hardware-in-the-Loop (HiL) testing, and experimental results.

Second, to ensure the long-term accuracy and utility of the DT, this thesis introduces a pioneering parameter estimation technique using Physics-Informed Neural Networks (PINNs) with unsupervised transfer learning. This method is a critical enabler for the DT, as it provides continuous, real-time updates of the converter's physical parameters. This ensures the DT model's accuracy, which is paramount for both enhancing control performance and for enabling predictive maintains. A key innovation of this PINN-based approach is its ability to operate using limited, low-frequency data without the need for precise synchronization with the converter's switching cycle, thereby overcoming a major practical barrier to deployment.

Finally, the scalability of the DTPC framework is demonstrated through its extension to DC microgrid applications. Collectively, the research presented in this thesis provides a validated pathway toward creating the next generation of intelligent and adaptive power electronic systems.

Date of Award	15 Nov 2025
Original language	English
Awarding Institution	University of Nottingham
Supervisor	Giampaolo Buticchi (Supervisor), Jing Li (Supervisor) & Alan Zhang (Supervisor)