Day-Ahead Natural Gas Demand Forecasting in Hourly Resolution

Abstract

Natural Gas (NG) demand forecasting is a research topic that starts to gather the attention of scholars, research institutions, utilities, retailers and other interested parties. Accurate predictions of future needs for NG can aid on the optimal management of NG resources. This manuscript examines the problem of day-ahead Natural Gas (NG) demand forecasting in hourly resolution. Various models of different type are trained and applied using data that correspond to the demand of a large region including urban, sub-urban and industrial loads. A series of scenarios are formed in order to investigate the influence of input selection on the day-ahead forecasting problem.