Electrical Energy Demand Forecasting Using Time Series Approach

Abstract

Load forecasting is a significant process of both smart grid and the deregulation power system. As the infrastructure of the smart grid is increasing day by day and also the adoption of deregulation principle in the present power system due to this the interest in electrical energy forecasting is increased. In this paper, time series models are used for predicting future electrical energy demand. In time series most common models called Autoregression Model (AR), Autoregression Moving Average (ARMA), Autoregression Integrated Moving Average (ARIMA) and Seasonal Autoregression Integrated Moving Average (SARIMA) is used for prediction. The accuracy of these methods is evaluated using Mean Absolute Percentage Error (MAPE) and Root Mean Squared Error (RMSE). Further, Paschim Gujarat Vij Company Limited (PGVCL, India), New York Independent System Operator (NYISO), and Energy Information Administration (EIA, USA) load data set are used for future demand forecasting. Using these three different data set, short-term load forecasting (STLF) and long-term load forecasting (LTLF) is carried out.