Note: This content is accessible to all versions of every browser. However, this browser does not seem to support current Web standards, preventing the display of our site's design details.


Complexity of First-Order Methods for Fast Embedded Model Predictive Control


D. Kouzoupis

Master Thesis, FS14 (10365)

Fast and efficient numerical methods for solving Quadratic Programming problems (QPs) in the area of Model Predictive Control (MPC) exist nowadays in so many variations that choosing the right solver for a specific application is no longer a trivial procedure. Although active-set and interior-point methods have monopolized the interest of control engineers over the past few decades, research in first-order methods has been recently revived and several algorithms based on Yurii Nesterovís fast gradient method (FGM) [34] have emerged. Some of the properties that contributed to this development is the simplicity of the algorithmic scheme, the guaranteed convergence rates and the ability to compute small to medium accuracy solutions in high sampling rates [23]. Following the growing interest in the field, the purpose of this Thesis is to provide a detailed overview of first-order solvers under a unifying framework and analyze their computational complexity in nonlinear MPC applications, where the generated QPs may comprise time-varying data. The results of the complexity analysis are then used as a guide to design an efficient MPC controller for an industrial example of ABB Corporate Research. iii


Type of Publication:

(12)Diploma/Master Thesis

P.J. Goulart

File Download:

Request a copy of this publication.
(Uses JavaScript)
% Autogenerated BibTeX entry
@PhdThesis { Xxx:2014:IFA_4850
Permanent link