How do extreme events impact the demand for energy?

Abstract

Extreme weather is considered one of the most dominating causes of disturbance for energy networks, especially electricity networks. Such weather events may affect the electrical grid infrastructure, But mainly affect the nature of consumption and demand for electricity on both industrial and domestic sides. The present work is based on historical weather and demand data to (a) understand how electricity demand changes under extreme event shock; (b) find correlations between changes in events weather parameters and changes in demand; (c) make predictions for estimating the future electricity changes under such extreme weather events. The aim of the work is to reach fair demand prediction accuracy of over 80% under the effect of heatwaves and hurricanes. Twelve machine learning regression models were used and tuned. The bestperforming model was Extra-trees for both cases. The heatwaves model score reached 14.6% root-mean-squared percentage error, While the hurricanes model score showed outstanding performance with 6.4% root-mean-squared percentage error.