Publication detail
NONLINEAR PREDICTIVE CONTROLLER WITH LOCAL APPROXIMATOR FOR ACTIVE MAGNETIC BEARING
PULCHART, J. BŘEZINA, T. TUREK, M.
Czech title
Nelineární predikující regulátor s lokálně váženým aproximátorem pro řížení aktivního magnetického ložiska
English title
NONLINEAR PREDICTIVE CONTROLLER WITH LOCAL APPROXIMATOR FOR ACTIVE MAGNETIC BEARING
Type
journal article - other
Language
en
Original abstract
Simulation study describes both behavior and learning of small active magnetic bearing control through nonlinear PD controller which uses parallel nonlinear predictive compensation, realized by local automatic approximator, which is presented further. Local approximator repeatedly uses the locally linear model based on algorithm of locally weighted regression. The goal of on‑line learning process is to minimize the mean square of control error. Obtained results are compared with results reached with nonlinear analytically designed PD controller. The controller used here is robust to high level noise and also able to work with relatively high delay.
Czech abstract
Simulační studie popisuje chování a učení řízení malého aktivného magnetického ložiska pomocí nelineárního PD regulátoru se zapojenou nelineární predikující kompenzací, realizovanou automatickým lokálním aproximátorem. Aproximátor používá lokální lineární modely založené na algoritmech lokálně vážené regrese. Jehož úkolem je, po naučení, minimalizovat regulační odchylku. Obdržený výsledek simulace řízení je porovnán s analyticky navrženým nelineárním PD regulátorem. Použitý regulátor je robustní pro vysoké hodnoty šumu a je schopen pracovat do relativně velkých zpoždění akčního zásahu.
English abstract
Simulation study describes both behavior and learning of small active magnetic bearing control through nonlinear PD controller which uses parallel nonlinear predictive compensation, realized by local automatic approximator, which is presented further. Local approximator repeatedly uses the locally linear model based on algorithm of locally weighted regression. The goal of on‑line learning process is to minimize the mean square of control error. Obtained results are compared with results reached with nonlinear analytically designed PD controller. The controller used here is robust to high level noise and also able to work with relatively high delay.
Keywords in English
Receptive Field Weighted Regression, nonlinear PD controller, nonlinear forward predictive compensator, prediction.
RIV year
2005
Released
20.09.2005
Publisher
Enginering Academy of the Czech Republic
Location
Jabloňová 5, 621 00 Brno
ISSN
1210-2717
Book
Mechatronics, robotics and biomechanics 2005
Journal
Inženýrská mechanika - Engineering Mechanics
Volume
2005
Number
13-3
Pages count
7
BIBTEX
@article{BUT42744,
author="Jaroslav {Pulchart} and Tomáš {Březina} and Milan {Turek},
title="NONLINEAR PREDICTIVE CONTROLLER WITH LOCAL APPROXIMATOR FOR ACTIVE MAGNETIC BEARING",
journal="Inženýrská mechanika - Engineering Mechanics",
booktitle="Mechatronics, robotics and biomechanics 2005",
year="2005",
volume="2005",
number="13-3",
month="September",
publisher="Enginering Academy of the Czech Republic",
address="Jabloňová 5, 621 00 Brno",
issn="1210-2717"
}