Title:  Sampled-data/continuous-time model free adaptive control: modeling, design, and robustness

Abstract:  In practical implementations, physical plants are typically controlled using either continuous-time or sampled-data control laws. Traditional model-free adaptive control (MFAC) methods are inherently designed for discrete-time systems, which often leads to a structural mismatch between the plant dynamics and the controller formulation. This lecture presents two novel MFAC design frameworks—one based on continuous-time control and the other on sampled-data control—both of which aim to better align the controller structure with the actual plant characteristics. Central to these frameworks are new dynamic linearization techniques that facilitate the systematic development of control and adaptation laws. By casting the resulting closed-loop dynamics as (weakly) interconnected nonlinear systems, we introduce novel Lyapunov-based robustness analysis approaches that explicitly account for external disturbances—and, in the sampled-data case, also for discretization errors. Furthermore, the theoretical analysis for the sampled-data MFAC rigorously quantifies the interplay among tracking error convergence, allowable ranges of adaptive parameters, and sufficiently small sampling periods. Finally, experimental validations of the proposed sampled-data MFAC are demonstrated through vehicle speed tracking tests and spacecraft attitude angle tracking applications.

Biography: Hao Yu is a Professor and Ph.D. Supervisor at Beijing Institute of Technology and was awarded the National Excellent Young Scientists Fund (Overseas) in 2022. He received his B.E. degree in 2013 and Ph.D. degree in 2018 from the School of Automation Science and Electrical Engineering, Beihang University. From 2019 to 2022, he conducted postdoctoral research in the Department of Electrical and Computer Engineering at the University of Alberta, Canada. His main research interests include networked control systems, event-triggered control, multi-agent systems, data-driven PID control, neural network adaptive control, and cyber-physical systems. He has published more than 40 academic papers as first or corresponding author, including over 10 papers in leading control journals such as IEEE Transactions on Automatic Control and Automatica.