Bridging Model Predictive Control and Robust Linear Control Theory
Buch in deiner Nähe kaufen
Beschreibung
Model predictive control (MPC) has been successfully applied for control in a wide variety of applications. It is an optimization based control algorithm consisting of a cost function which includes output predictions over a certain prediction and a control horizon as well as penalty matrices on control errors and inputs. Suitable tuning of the horizons and the penalty matrices is essential for an effective control with MPC. Despite some existing guidelines on the choice of tuning parameters, very little is understood on how to choose them effectively such that a reasonable trade-off between the performance and the robustness is achieved. This lack of understanding poses a fundamental challenge for the successful design of MPC. The aim of this thesis is to develop an approach for the systematic determination of the tuning parameters.
To accomplish the goal, an underlying idea is to treat MPC within the framework of linear control theory. Using the latter, the tuning problem is made systematic by considering MPC as a classical controller design problem.
The tuning parameters are determined by solving two-step convex optimization problems where the objective is to minimize the difference between the true and the desired closed-loop performances. The tuning procedure is applied to SISO and MIMO systems. The said tuning approach has a potential drawback that the robust stability cannot be guaranteed. Therefore, the tuning approach is revised to approximate the true open-loop performance with an achievable open-loop performance where the latter is determined from an achievable closed-loop performance.
The revised tuning approach is applied for the computation of robust MPC and is successfully implemented on three case study examples.
Autor*in
Guarang Shah
Themen in »Bridging Model Predictive Control and Robust Linear Control Theory«
Linear MPC Linear Model Predictive Control MPC Tuning linear control theory