Urban electric load forecasting with mobile phone location data

Abstract

In recent years, electrical load forecasting has received continuous research efforts aiming to improve the short-term prediction accuracy of local energy demands. However, current methods are not able to take explicitly into account the dynamic spatial population distribution over the course of a day, which is particularly relevant in dense urban areas. In this paper, we harness society-wide mobile phone data to map the time-varying population distribution in the Trentino region, Italy, and to use these insights for a novel electrical load forecasting method. Our results demonstrate that the integration of aggregated mobile phone data yields compelling forecast models.