للكاتبين :
A.Y. Haikal, M.S.M. KSASY, S.F. SARAYA, F.F. AREED
Computers & Systems Dept., Faculty of Engineering, Mansoura University
ABSTRACT: –
This work is concerned with studying an effective evolutionary algorithm based on neuron level
beside blueprint level search algorithm. This algorithm maintains diversity in the population due to
the evolution of partial solutions (neurons) instead of full solutions (networks). Each neuron
chromosome is composed of the incoming & outgoing connections & weights of the hidden-layer
neurons. Algorithm was investigated on an inverted pendulum system which is a common classical
control problem. It is a suitable process to test prototype controllers due to its high non-linearities and
lack of stability. The developed controller was proved to be both robust and of high performance for
controlling highly nonlinear and unstable plant. Also, after the investigation of the developed
controller’s ability to generalize, it displayed a high level of robustness to parameter uncertainty in
the model.