Résumé
The general context of this thesis concerns the integration of RESs (Renewable Energy Sources) into smart MGs (micro-grids) for buildings in order to support the continuous growth of buildings’ electricity demands. However, their intermittent nature and unpredictable variability represent the main challenge of their efficient and seamless integration into buildings. Energy storage systems are considered among the most promising technologies that could balance RESs production with buildings’ energy consumption. Electrochemical (batteries) storage systems are the most deployed in buildings. This is due to their multiple advantages, mainly modularity, cleanliness, and high efficiency. However, the unpredictable and discontinuous nature of the power production and consumption make the power management in MG systems a difficult task. Therefore, intelligent control strategies are required for efficient energy management in MG systems.This thesis focuses on the development and deployment of an intelligent and predictive control strategy for energy balance in MG systems. A predictive control approach, named MAPCASTE (Measure, Analyze, Plan, ForeCAST, and Execute), is developed and deployed in real-sitting scenarios. Mainly, MPC (Model Predictive Control) and GPC (Generalized Predictive Control) have been investigated in order to carry out the proposed MAPSASTE. This later was deployed and evaluated by assessing its effectiveness for energy management in MG systems. In particular, modeling, simulation, experimentation, and performance assessment of the deployed MG system, together with the developed control approach, have been performed. Experimentations have been conducted using a real MG platform, which was deployed in the frame of two research & development projects.