الفهرس 
Chabter 1 : introductionOnly 14 pages are availabe for public view
 |  | from | 186 |  |  |
| | | from | 186 | | |
Abstract
Neural networks have been received an increasing attention in recent few years because they show a great potential for identify ,model and control systems that are difficult or impossible to model using traditional techniques neural net works which have the ability of learning sophisticated nonlinear systems relation ships provide promising mean for modeling nonlinear systems neural networks based controller or neuro-controller achieve good performance when the system parameters are poorly or completely unknown if just we have an input – out put data collected from the system.
In this thesis neural networks are first analyzed their function approximation capabilities are investigated two learning methods are discussed and compared back propagation and marquee dt-levenberg methods system identification based on neural networks procedure and program are shown and examined by using correlation based model validity tests also tow algorithms for system control based on neural networks are investigated and compared direct inverse control (DIC) and internal models CONTROL (IMC)