A0552
Title: Forecasting electricity consumption at household level
Authors: Fatima Fahs - University of Strasbourg (France) [presenting]
Frederic Bertrand - Universite de technologie de Troyes (France)
Myriam Maumy-Bertrand - University of Technology of Troyes (France)
Abstract: National-scale electrical load forecasting has been a topic of research for decades. A variety of models and techniques have been successfully developed to accomplish this task. The recent massive deployment of smart meters at the household scale has stimulated research on electrical load forecasting at this scale, both in academia and in industry. An accurate forecast of load at the scale of an individual household is beneficial to both electricity providers and consumers. Electricity providers rely heavily on household-scale load forecasts to improve the efficiency of smart grid applications, such as smart home energy management systems and demand response applications. Thus, by providing their customers with advanced services, they can optimize their electricity consumption and reduce their electricity bills. Due to the high volatility of load curves at the household level, forecasting load at this scale remains a challenge for researchers, especially since the research published on this subject so far does not meet the real-world challenge. In our study, we propose daily forecast models of half-hourly electrical loads at the household level. The performance of the models is evaluated on disparate real load curves of the residential and tertiary sectors provided by a French electricity company. Also, we provide a forecasting approach for the most volatile load curves. Our study takes into account the industrial constraints to propose an industrially viable approach.
