Title :
Approximation and control of systems using a neural net
Author :
Ermish, M. ; Nouri-Moghadam, M.
Author_Institution :
Dept. of Math., Penn State Univ., PA, USA
Abstract :
The use of multilayer neural networks for approximating linear and nonlinear systems is demonstrated. The gradient method is discussed in detail. Single-input/single-output and double-input/single-output systems are considered, and gradient (backpropagation) methods are used to adjust the parameters of a three-layer neural network in order to optimize a performance function. The results of simulations for the above systems are analyzed, and appropriate graphs that verify the theoretical results are included. The approach is also used for approximating the optimal control of vibrating beams
Keywords :
approximation theory; backpropagation; discrete systems; multilayer perceptrons; network parameters; neurocontrollers; nonlinear control systems; optimal control; simulation; vibration control; backpropagation; gradient method; multilayer neural networks; optimal control; performance function; simulations; vibrating beams; Analytical models; Backpropagation; Control systems; Gradient methods; Linear approximation; Multi-layer neural network; Neural networks; Nonlinear systems; Optimal control; Optimization methods;
Conference_Titel :
System Theory, 1993. Proceedings SSST '93., Twenty-Fifth Southeastern Symposium on
Conference_Location :
Tuscaloosa, AL
Print_ISBN :
0-8186-3560-6
DOI :
10.1109/SSST.1993.526520
