`a`
Discrete and Continuous Dynamical Systems - Series A (DCDS-A)
 

Robustness of performance and stability for multistep and updated multistep MPC schemes

Pages: 4385 - 4414, Volume 35, Issue 9, September 2015      doi:10.3934/dcds.2015.35.4385

 
       Abstract        References        Full Text (569.8K)       Related Articles       

Lars Grüne - University of Bayreuth, Chair of Applied Mathematics, Universitätsstraße 30, 95440 Bayreuth, Germany (email)
Vryan Gil Palma - University of Bayreuth, Chair of Applied Mathematics, Universitätsstraße 30, 95440 Bayreuth, Germany (email)

Abstract: We consider a model predictive control approach to approximate the solution of infinite horizon optimal control problems for perturbed nonlinear discrete time systems. By reducing the number of re-optimizations, the computational load can be lowered considerably at the expense of reduced robustness of the closed-loop solution against perturbations. In this paper, we propose and analyze an update strategy based on re-optimizations on shrinking horizons which is computationally less expensive than that based on full horizon re-optimization, and at the same time allowing for rigorously quantifiable robust performance estimates.

Keywords:  Optimal control, model predictive control, robustness against perturbations.
Mathematics Subject Classification:  Primary: 49K20, 93B52, 49M37; Secondary: 49L20.

Received: May 2014;      Revised: August 2014;      Available Online: April 2015.

 References