للكاتبين :
A.Y. HAIKAL, M . S .M. KSASY, S. F . SARAYA, F . F . AREED
Computers & Systems Dept., Faculty of Engineering, Mansoura University
ABSTRACT:-
This article investigates the performance of a modified reinforcement evolutionary algorithm
when applied to control the pH process in a continuous stirred tank reactor. This system exhibits
highly nonlinear behavior and has very high gains around the electroneutrality value. The
proposed algorithm works on two levels, neuron-level and blue prints networks-level. The
present algorithm differs from standard evolutionary algorithms in that it evolves a population of
neurons to be combined in a later stage and form the hidden layer instead of evolving complete
networks directly. Moreover, these neurons are combined through different subpopulations. The
formed feed-forward networks are then evaluated on a given problem. The neurocontroller
developed was found to be robust and presents high performance during set point changes with
no oscillations and it proves its robustness through its ability to reject sudden disturbances
resulted from changing acid flow rate.