Demand Forecasting in Deregulated Electricity Markets

Abstract

day ahead demand forecasting is essential for the efficient operation of electricity companies in the competitive electricity markets. Both the power producers and consumer needs single compact and robust demand forecasting tool for the efficient power system planning and execution. This research work proposes a day ahead short term demand forecasting for the competitive electricity markets using Artificial Neural Networks (ANNs). Historical demand data are collected for the month of January 2014 from PJM electricity markets. The work proposes the approach to reduce prediction error for electricity demands and aims to enhance the accuracy of next day electricity demand forecasting. Two types of demand forecasting models: classical forecasting and correlation forecasting models are proposed, explained and checked against each other. Proposed models are applied on real world case, PJM electricity markets for forecasting the demand on weekly working day, weekly off day and weekly middle day. The Mean Absolute Percentage Error (MAPE) for the two proposed models in the three respective cases is evaluated and analyzed. Results present that with all respects a day ahead demand forecasting through the correlation model are best and suitable for PJM electricity markets and produce less error with comparison of other classical models.