A feedforward neural network and its application to system indentification and control

 The aim of this thesis is to study a feedforward neural network and its application to system identification and control. 

Attention is focused firstly on feedforward neural networks and their weight adaptation algorithms. A new class of weight adaptation learning algorithms are introduced based on the sliding mode concept. The effectiveness of the new class of algorithms are studied and simulations are conducted to present their performance. 

Second part of this thesis deals with the application of the feedforward neural network with the developed learning algorithms. Two classes of problems are chosen to test the suitability of the feedforward neural network with proposed adaptation learning algorithms. The first problem is dynamic system identification and the other is dynamic system control. Results are presented in this thesis show the effectiveness of the feedforward neural network with the proposed learning algorithms in system identification and control.