Smart thermostats provide heating and/or cooling by sensing actual (or predicted) thermal needs from occupants in the different rooms, acknowledging the current (or predicted) thermal conditions of the house, assessing the thermal response of systems (mechanical and enclosure) to current and predicted occupancy demands and weather, and responding promptly and smoothly. The fundamental goal of smart thermostat is to provide just enough heating/cooling to meet occupants thermal demands, while minimizing the use of energy.
Ultimately, the smartness of a thermostat depends on the responsiveness of the environmental systems. For example a smart thermostat can be coupled with smart windows that sense the environment, open/close automatically and/or inform occupants when to open windows for passive cooling, natural ventilation, and energy saving.
Research question: how to consider human behaviors and quantify human-building interactions?
Methods:
- Collect, process, and analyze thermostatic field data and possibly relevant indoor environmental and occupancy data
- Develop mechanistic dynamic simulation models to explore the response (time-constant) of the building
- Develop and test a controls model that can either be: simplified (i.e. principle parameters-identified) mechanistic (e.g thermal network), statistical (black-box), artificial intelligence (AI) based, or hybrid.
- Validate controls model using field data and/or by cross-validating models
References
- Shann M., Alan A., Seuker S., Constanza E., and Ramchurn S. (2017). Save Money or Feel Cozy?: A Field Experiment Evaluation of a Smart Thermostat that Learns Heating Preferences, Proceeding AAMAS ’17 Proceedings of the 16th Conference on Autonomous Agents and MultiAgent Systems, pp. 1008-1016, São Paulo, Brazil — May 08 – 12, 2017
. Learning from heating preferences smart thermostats