Title :
Optimal feedback control of nonlinear systems using a neural network
Author_Institution :
Dept. of Electr. Eng., Ottawa Univ., Ont., Canada
Abstract :
A novel approach that simplifies the problem of finding the optimum feedback control law using a multilayer feed-forward neural network is presented. The network can be trained using a weight adjustment algorithm derived directly from the integral form of Pontryagin´s minimum principle to minimize a certain cost function. The capability of the proposed approach is demonstrated by controlling the attitude of a rigid body satellite in geosynchronous orbit
Keywords :
cost optimal control; feedback; feedforward neural nets; minimum principle; multilayer perceptrons; neurocontrollers; nonlinear control systems; Pontryagin´s minimum principle; cost function minimization; geosynchronous orbit; multilayer feed-forward neural network; nonlinear systems; optimal feedback control; rigid body satellite attitude control; weight adjustment algorithm; Adaptive control; Cost function; Feedback control; Feedforward systems; Neural networks; Nonlinear equations; Nonlinear systems; Open loop systems; Optimal control; Satellites;
Conference_Titel :
Electrical and Computer Engineering, 1995. Canadian Conference on
Conference_Location :
Montreal, Que.
Print_ISBN :
0-7803-2766-7
DOI :
10.1109/CCECE.1995.526306
