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.

  

Convex optimization tailored for real-time Model Predictive Control

Author(s):

Cl. Wiltsche
Conference/Journal:

Semester Thesis,HS 10 (10064)
Abstract:

In recent years, it has been shown that the computation times for solving an MPC problem can be pushed into a range where online optimization becomes a reasonable alternative for the control of high-speed systems. Significant reduction of the computational complexity can be achieved by exploiting the particular structure and sparsity of the optimization problem using tailored solvers. In this semester project, a code generator has been developed that produces code for interior point methods for MPC. This has been used to compare the performance of computing the search direction either via a factorization or a gradient method. It has been shown that the pure C implementation using gradient methods exhibits linear complexity in the prediction horizon. One motivation for this was to be able to decouple a controller from having to provide a Fortran compiler. The code generator allows the user to conveniently select between the factorization or the gradient method, which allows to efficiently generate code that is optimized for a specific MPC problem.

Year:

2011
Type of Publication:

(13)Semester/Bachelor Thesis
Supervisor:

C.N. Jones

No Files for download available.
% Autogenerated BibTeX entry
@PhdThesis { Xxx:2011:IFA_4122
}
Permanent link