Designing building automation systems using evolutionary algorithms with semi-directed variations

In the building automation domain, many prefabricated devices from different manufacturers available in the market realize building automation functions by preprogrammed software components. For given design requirements, the existence of a high number of devices that realize the required functions leads to a combinatorial explosion of design alternatives at different price and quality levels. Finding optimal design alternatives is a hard problem to which we approach with a multi-objective evolutionary algorithm. By integrating problem-specific knowledge into variation operations, a promisingly high optimization performance can be achieved. To realize this, diverse variation operations related to goals are defined upon a classification for the exploration and convergence behavior, and applied in different strategies.