Encodings for Evolutionary Algorithms in smart buildings with energy management systems

In energy systems, the transition from traditional, centralized architecture and controllable generation to an ever more decentralized and volatile generation due to an increasing use of renewable energy sources arises new challenges for the management and balancing of the electricity grid. These can be met through energy management systems (EMS) that enable flexible consumption and production of energy on the demand side of the grid. The EMS for smart buildings that is used within this paper allows for the integration of a multitude of devices through an architectural approach which is similar to “plug-and-play”. These devices can then be optimized to a flexible load shape by an Evolutionary Algorithm (EA). The differentiated optimization capabilities of the devices require adequate encoding schemes. Such schemes are the major contribution of this paper. The aptitude of these encodings is shown and validated through the simulation of smart buildings with different configurations, both concerning quantitative and qualitative benefits to be achieved according to energy systems' transition and users' objectives.