Load Forecasting Using Artificial Neural Network

Abstract

Load forecasting on short term basis is formulated in the framework of Artificial Neural Network (ANN) model. The maltilayered perception model using back propagation algorithm makes possible to train the ANN with training patterns. The training patterns (past data) can act as input in various ways which are explored and reported here. The trained network is tested on the MSEB data, and the results are discussed. The network can be trained in a dynamic way by using the most recent data.