A Predictive Approach to Intelligent Building Systems Control

Abstract

This keynote presents the implementation of a novel predictive approach to intelligent building systems control. The implementation specifically pertains to the utilization of passive cooling in buildings. Thereby, numeric simulation is deployed as an integral part of the control logic to predict future implications of alternative control options (alternative positions of windows, shades, etc.) and identify the best performing control option. A genetic algorithm was developed to generate a manageable set of alternative options from the corpus of all possible control actions at any given time. Five rooms in two office buildings in Austria were used to test this method. The paper describes the approach and implementation in detail and presents the results.