DocumentCode :
707003
Title :
Ahybridneural-decouplingpole placement controller and its application
Author :
Henriques, J. ; Dourado, A.
Author_Institution :
Dept. de Eng. Inf., Univ. of Coimbra, Coimbra, Portugal
fYear :
1999
fDate :
Aug. 31 1999-Sept. 3 1999
Firstpage :
3946
Lastpage :
3951
Abstract :
A hybrid control architecture is proposed integrating recurrent, dynamic neural networks into the pole placement context. The neural network topology involves a modified recurrent Elman network to capture the dynamics of the plant to be controlled, being the learning phase implemented on-line using a truncated backpropagation through time algorithm. At each time step the neural model, modelling a general non-linear state space system, is linearized to produce a discrete linear time varying state space model. Once the neural model is linearised some well-established standard linear control strategies can be applied. In this work the design of a decoupling pole placement controller is considered at each instant, which combined with the on-line learning of the network results in a self-tuning adaptive control scheme. Experimental results collected from a laboratory three tank system confirm the viability and cffcctivcncss of the proposed methodology.
Keywords :
adaptive control; backpropagation; linear systems; neurocontrollers; nonlinear control systems; pole assignment; recurrent neural nets; self-adjusting systems; state-space methods; discrete linear time varying state space model; dynamic neural networks; hybrid neural-decoupling pole placement controller architecture; modified recurrent Elman network; neural network topology; nonlinear state space system; plant dynamics; pole placement context; self-tuning adaptive control scheme; standard linear control strategies; time algorithm; truncated backpropagation; Adaptation models; Aerospace electronics; Control systems; Heuristic algorithms; Mathematical model; Neural networks; Standards; Hybrid methods; decoupling; multivariable adaptive control; pole placement; recurrent neural networks;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Control Conference (ECC), 1999 European
Conference_Location :
Karlsruhe
Print_ISBN :
978-3-9524173-5-5
Type :
conf
Filename :
7099948
Link To Document :
بازگشت