Abstract
In this paper, a novel approach based on the combination of the Simulating Annealing (SA) algorithm, as an optimization tool, and Artificial Neural Networks (ANN), as an adaptation technique, with dominant eigenvalue shift to design an optimized self-tuned Proportional-Integral-Derivative (PID) controller that may overcome difficulties faced when a change in system parameters occurs. The proposed approach has been implemented as a voltage regulator for a synchronous generator connected to an infinite-bus power system. The optimization search is based on a suitable objective function. ANN is trained off-line for several operations conditions and then employed for fast online prediction of the system model and controller gains. To demonstrate the effectiveness of the obtained controller, the synchronous generator, equipped with such optimized tuned regulator, is tested under different operating conditions and parameter changes. Its robustness is shown through comparison with the well-known IEEE voltage regulator and the optimization process via Ziegler-Nichols technique. The results show the capability of the proposed controller to enhance well the system performances.