Online Learning Neural Network based PSS with Adaptive Training Parameters
Author(s): Pinak Tulpule, Ali Feliachi
Reference: ProceedingsIEEE PESGeneral Meeting 2007, Tampa, Florida, June 2007
Abstract: This paper provides a new method to improve power system stability using recurrent neural network with adaptive training parameters. Power system generators are equipped with automatic voltage regulator, power system stabilizer, and governor to control and stabilize the system. The controller parameters are tuned using mathematical methods, or heuristic search methods such as genetic algorithm. Therefore these control parameters are often fixed and are set for particular system configurations or operating points. Artificial neural network can be tuned for changing system conditions and thus provide better control. Artificial neural network is used in this paper in parallel with the existing PSS to effectively damp the oscillations and improve overall system performance. Online training method is employed with multilayer recurrent neural network. Training is based on back propagation with adaptive training parameters. This controller is tested on two different systems and simulation results are presented to illustrate the proposed approach.
Keywords: power systems, transient stability, neural networks,