Comparison of feature selection techniques for ANN-based voltage estimation

Fast estimation of bus voltage magnitude is essential for security monitori ng and analysis of power system. An approach based on a parallel self-organ ising hierarchical neural network (PSHNN) is proposed to estimate bus volta ge magnitudes at all the PQ buses of a power system in an efficient manner. PSHNN is a multi-stage neural network in which stages operate in parallel rather than in series during testing. The revised back-propagation algorith m is used for learning input non-linearities along with forward-backward tr aining of stage neural networks. A method based on Euclidean distance clust ering is proposed for feature selection. Effectiveness of the proposed meth od is compared with two existing methods of feature-selection entropy based and angular distance based clustering methods for bus voltage magnitude es timation at different loading conditions in the IEEE 30-bus system and a pr actical 75-bus Indian system. The PSHNN based on Euclidean distance based c lustering method is found to be superior in terms of training time and erro r performance. (C) 2000 Elsevier Science S.A. All rights reserved.