On-line Parameter Tuning of Model Predictive Control

0363300 - ÚTIA 2012 RIV IT eng C - Konferenční příspěvek (zahraniční konf.)
Belda, Květoslav
On-line Parameter Tuning of Model Predictive Control.
Proceedings of the 18th IFAC World Congress. Milano: IFAC - International Fedaration of Automatic Control, 2011 - (Bittanti; Cenedese; Zampieri), s. 5489-5494. ISBN 978-3-902661-93-7.
[18th IFAC World Congress. Milano (IT), 28.08.2011-02.09.2011]
Grant CEP: GA ČR(CZ) GAP102/11/0437
Výzkumný záměr: CEZ:AV0Z10750506
Klíčová slova: Model Predictive Control * Parameter Tuning * LQ Control * Probality Calculus
Kód oboru RIV: BC - Teorie a systémy řízení
http://library.utia.cas.cz/separaty/2011/AS/belda-on-line parameter tuning of model predictive control.pdf

Model Predictive Control is a powerful control approach suitable for industrial applications due to its simplicity and flexibility. The aim of the paper is to introduce principal points for on-line tuning and automatic set-up of predictive control parameters. A proposed way of the tuning arises from the fact that in all control tasks, there is a certain degree of uncertainty with respect to the controlled system including its neighborhood. The strategy takes advantage of an analogy of the discrete Predictive Control and Linear Quadratic Control (LQ Control) connected with the probability calculus. The parameter tuning is demonstrated on multidimensional robotic system.
Trvalý link: http://hdl.handle.net/11104/0199279