(Big) data analytics in smart grids

Prediction of building energy consumption is a fundamental problem in the smart grid context. Unprecedented high volumes of data and information are available with the upward growth of the smart metering infrastructure. Therefore, we develop two deep learning methods. Firstly, the demand forecasting problem was solved at low aggregation levels (i.e. 1900 buildings) using factored conditional restricted Boltzmann machine. Secondly, we developed an unsupervised energy prediction method using reinforcement cross-building transfer able to accurately estimate the energy based on the information available in the neighborhood. Both methods have been successfully validated on real-world databases.