Algorithms and methodologies for decision support in energy efficiency on buildings

Dissertação apresentada na faculdade de Ciências e Tecnologia da Universidade Nova de Lisboa para a obtenção do grau de Mestre em Engenharia Electrotécnica e de Computadores

Buildings worldwide account for approximately 40 percent of the global energy consumption and the resulting carbon footprint significantly exceeds those of all transportation combined. However, large and attractive opportunities to reduce energy use in buildings exist today. To reach ambitious energy efficiency goals, the building sector must undergo through technological innovation, informed customer choices, and smart business decisions. Existing building simulation tools provide users with key building performance indicators,such as energy use and demand. However, these tools do not deal with activities performed by building occupants and with the resulting utilization of spaces. At best, they rely on assumptions referring to human behavior. As a result, energy prediction often does not represent the real building utilization. Therefore, it is assumed that user behavior is one of the most important input parameter influencing the results of building performance simulations. A methodology for constructing an energy consumption model that reflects the human behavior dynamics and occupancy patterns within a building is presented. This research will provide a possible methodology for the pillars of future work in modeling the building usage under real patterns of utilization. A simulator has been developed from a model where both human behavior and building have been incorporated. Simulations have been performed to test different behavioral situations where the developed models and algorithms have been applied for prediction purposes. The proposed methodologies focus on the applicability of a rule-based expert system to support the simulator and stochastic modeling. The building’s occupant behavior is modeled with a hidden Markov model and the building’s spaces are described as Markov chains.