Customizable Energy Management in Smart Buildings Using Evolutionary Algorithms

Various changes in energy production and consumption lead to new challenges for design and control mechanisms of the energy system. In particular, the intermittent nature of power generation from renewables asks for significantly increased load flexibility to support local balancing of energy demand and supply. This paper focuses on a flexible, generic energy management system for Smart Buildings in real-world applications, which is already in use in households and office buildings. The major contribution is the design of a “plug-and-play”-type Evolutionary Algorithm for optimizing distributed generation, storage and consumption using a sub-problem based approach. Relevant power consuming or producing components identify themselves as sub-problems by providing an abstract specification of their genotype, an evaluation function and a back transformation from an optimized genotype to specific control commands. The generic optimization respects technical constraints as well as external signals like variable energy tariffs. The relevance of this approach to energy optimization is evaluated in different scenarios. Results show significant improvements of self-consumption rates and reductions of energy costs.