ICT for automated forecasting of electrical power consumption: A case study in Maputo

Accurate short term load forecasting is crucial for efficient operations planning of electrical power systems. We present a model for automatic forecasting of the short term (24 hours) electrical power consumption in Maputo, Mozambique. The proposed model is based on analysis of historical records of power consumption combined with information about additional factors that influence the consumption. The data is clustered into segments with the objective of identifying similar consumption patterns. These consumption patterns are then correlated with weather conditions and used to construct an automated prediction model for load forecasting. Today these forecasts are made manually by experts at Electricidade de Moçambique (the local power company) using conventional methods. The automated prediction model that was developed in this project presents an accurate and consistent complement to manual prediction and is currently being evaluated for the possibility of augmenting the manual forecasts with additional information.