Short Term Load Forecasting Using BPN and RBF Network

Abstract

Simple neural network based short term load forecaster is designed which predicts load
values to be obtained beforehand. The neural network based short term load forecaster has
two modules (1). Back Propagation Network (BPN) and (2) Radial Basis Function Network
(RBF). The inputs used were the actual hourly load demand for the full day (24 hours) and
the outputs obtained were the predicted hourly load demand for the next day. The number of
inputs is 25 while the number of hidden layer neurons is varied for different performance of
the network and the output layer has 24 neurons. The results obtained from two different
approaches are compared and accuracy of neural network is reported better. Also, the
network has been trained over one week and an absolute mean error of 2.64% was achieved
when the trained network was tested on one week’s data. Short-term hourly load forecasting
is predicted using Matlab R2010a toolbox.