Quantile Regression Model for Peak Load Demand Forecasting with Approximation by Triangular Distribution to Avoid Blackouts

• Main Authors: Niematallah Elamin, Mototsugu Fukushige
• Format: Article
• Language: English
• Published: EconJournals 2018-09-01
• Series: International Journal of Energy Economics and Policy
• Online Access: https://www.econjournals.com/index.php/ijeep/article/view/6742

<p dir="RTL" align="right">Peak load demand forecasting is a key exercise undertaken to avoid system failure and power blackouts. In this paper, the next day’s peak load demand is forecasted. The challenge is to estimate a model that is capable of preventing underprediction of the peak load demand: in other words, a model that is competent in forecasting the upper bound of the peak demand to avoid the risk of power blackouts. First, quantile regression is performed to generate forecasts of the daily peak load demand. Then, peak demand forecasts are locally approximated by triangular distribution to generate the upper bound of the peak demand. The forecasted upper bounds are compared with the actual electricity demand. The proposed method succeeds in avoiding underprediction of the peak load demand and thus the risk of power blackouts.</p><p dir="RTL" align="right"><strong>Keywords</strong><strong>: </strong>Electricity peak demand, Quantile regression, Triangular distribution, Blackouts.<strong></strong></p><p dir="RTL" align="right"><strong>JEL Classifications</strong>: Q47, C21</p>