Identification of Suitable Learning Algorithm for Neural Network based On-Line Economic Load Dispatch Problem

Abstract

This paper identifies suitable learning algorithm for neural network based Economic Load Dispatch (ELD). ELD is an important problem in real time power system planning and Operation. The conventional methods used for economic load dispatch are iterative techniques & takes longer time for computation. Neural Network (NN) provides an alternate solution for on-line Load Dispatch. The on-line Load Dispatch requires the NN model to be accurate, simple and structurally compact to ensure faster execution time for effective load dispatch. This in turn to a large extent depends on the type of Neural Learning algorithms used to train the Neural Architecture. The feed forward neural architecture is trained off-line using three types of learning algorithms to solve economic load dispatch problem. Their performance is compared in terms of accuracy, computational complexity and structural compactness. The results are validated for IEEE 26 Bus system. The promising results obtained are presented.