ART artificial neural networks based adaptive phase selector

This paper introduces a new phase selector based on adaptive resonance theory (ART). Because conventional phase selector cannot adapt dynamically to the power system operating conditions, it presents different characters under different power system conditions. To overcome the disadvantage, an adaptive phase selector, which utilizes artificial neural network based on ART, is designed. ART based neural network (ARTNN) has some advantages such as no local extremum, quickly convergence and so on. Therefore, the proposed ARTNN based phase selector has better performances compared with other neural networks based phase selector, and the new selector can adapt dynamically to the varying power system operation conditions. Furthermore, the phase selector can be trained and learned on-line. A lot of EMTP simulations and experimental field data tests have illustrated the phase selector's correctness and effectiveness. (C) 2005 Elsevier B.V. All rights reserved.