Build a system for forecasting the electrical load demand in Baghdad

Abstract

This research studies how to build a dynamic system for forecasting the electrical demand in Baghdad city by comparing between statistical methods in time series analysis such as Seasonal Auto Regressive Integrated Moving Average model (SARIMA),Transfer Function Model with single input-single (TFM-SISO) output to analyze the data for their dynamic structure and Data Mining techniques in prediction in Artificial Neural Networks (ANN) such as MLP-NN With Sliding Windows Model and the Non-Linear Auto regressive with exogenous inputnetwork (NARX Network) , which studies the dynamic relationship between electricity consumption and its relevant variables exogenous variable such as temperature, the Weekly data from January 2007 to December 2014 for all-electric residences in Baghdad are used for this study. Depending on the automated system that built by using (Visual C#, Matlap(The results showed superiority of the non-linear Auto regressive with exogenous input network (NARX Network)) by using some error criterion .