A computationally efficient approach for robust gain-scheduled output-feedback LQR design for large-scale systems

Paper in proceeding, 2020

This paper proposes a novel and simple control design procedure for sub-optimal robust gain-scheduled (GS) output-feedback linear quadratic regulator (LQR) design for large-scale uncertain linear parameter-varying (LPV) systems. First, we introduce a simple and practical technique to convexify the controller design problem in the scheduled parameters. Then, we propose a computationally efficient iterative Newton-based approach for gain-scheduled output-feedback LQR design. Next, we propose a simple modification to the proposed algorithm to design robust GS controllers. Finally, the proposed algorithm is applied for air management and fueling strategy of diesel engines, where the designed robust GS proportional-integral-derivative (PID) controller is validated on a benchmark model using real-world road profile data.