Number: VEGA 1/0095/11
Principal investigator: M. Kvasnica
Keywords: model predictive control, real-time control
The project is aimed at conducting research in the area of real-time implementation of Model Predictive Control
(MPC) using hardware control platforms with limited computational power and constrained memory storage. Such
constraints are typical for a broad class of industrial control systems, including, but not limited to, digital signal
processors of programmable logic controllers. Therefore the main focus of the project is to develop novel
theoretical approaches aimed at reducing the computational demands of MPC implementation in real time and to
provide unique software tools for design, analysis, verification and implementation of predictive controllers. The
main goal is to achieve faster and cheaper implementation of MPC on industrial control systems. Results of the
projects will be verified on a large number of real-life control systems and published in international journals.
The project is aimed at research in the area of implementation of Model Predictive Control using industrial control
systems, which, typically, have limited computational power and low memory storage. A successful solution to
such a complex problem involves answering the following main questions:
- How to find analytic solutions of MPC problems?
- How to simplify such a solution and reduce its memory footprint?
- How to evaluate such an analytic solution using the minimal number of computational operations?
- How to implement the solutions using various different control platforms?
- How to disseminate the results of the research in form of easy-to-use software packages?
It is up to the particulars goals of the project to answer each and every of these questions. The individual
problems will be solved in such a way that the results are applicable to a large variety of systems and control
platforms. In particular, the project will focus on developing novel MPC algorithms for the class of linear, hybrid,
fuzzy and nonlinear predictions models. Subsequently, the control algorithms will be experimentally verified on a
broad class of control systems including PLCs, DSPs and FPGAs. All theoretical results obtained in the course of
the project will be disseminated in a form of freely available software packages.