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.

  

Numerical Study on Real-Time Model Predictive Control for Piece-Wise Affine Systems

Author(s):

G. König
Conference/Journal:

Semester Thesis, FS15 (10434)
Abstract:

Currently, optimal control problems of systems with piecewise affine dynamics are reformulated to mixed-integer quadratic programs and are solved using branch-and-bound methods. Due to limited memory and computational power these solvers cannot typically be deployed on embedded platforms. Researchers at the Automatic Control Laboratory proposed a new method to efficiently solve these problems. An alternative reformulation gives affine equality constraints and independent stage-wise non-convex constraints, such that an operator splitting first-order method that performs euclidean projection computations on non-convex sets can solve the problem. The aim of this project was to investigate the proposed method regarding convergence, solve time and quality of a solution, i. e. can the method be tuned such that it converges for every problem and initial condition, how much faster is the method compared to a state-of-the-art solver and whether the obtained solution is a global or only local optimum. These questions are answered for a benchmark set of problems.

Supervisors: Juan Jerez, Alexander Domahidi, Manfred Morari

Year:

2015
Type of Publication:

(13)Semester/Bachelor Thesis
Supervisor:



File Download:

Request a copy of this publication.
(Uses JavaScript)
% Autogenerated BibTeX entry
@PhdThesis { Xxx:2015:IFA_5214
}
Permanent link