Title :
Whether and when the conventional controllers may operate well with nonlinear plants
Author :
Gessing, Ryszard
Author_Institution :
Inst. of Autom. Control, Silesian Univ. of Technol., Gliwice, Poland
Abstract :
It is shown that for strongly nonlinear plants taken from MATLAB Neural Network (NN) Control Systems Demos, the conventional controllers PD and P, which fulfill the demands of formulated in the paper the Criterion RD1 (Relative Degree = 1), are significantly better then NN controllers appearing in these Demos. Moreover the conventional controllers operate well for significantly larger changes of the reference signal without the need of their re-tuning.
Keywords :
neurocontrollers; nonlinear control systems; MATLAB neural network; NN controllers; control systems demos; nonlinear plants; Artificial neural networks; MATLAB; Mathematical model; Nonlinear dynamical systems; PD control; Training;
Conference_Titel :
Control & Automation (MED), 2010 18th Mediterranean Conference on
Conference_Location :
Marrakech
Print_ISBN :
978-1-4244-8091-3
DOI :
10.1109/MED.2010.5547666
