Abstract
Personal Comfort Models (PCM) build on state-of-the-art methods to predict individuals’ thermal comfort responses based on occupant input and indoor environment measurements. Such models use machine learning and internet of things (IoT) to learn personal comfort requirements and integrate them into control systems for indoor environments. However, it is a challenging task to have consistent feedback from individuals, which is normally obtained from thermal comfort surveys. In this study, a PCM approach was tested, where the effects of thermal adaptation on occupants’ thermal responses were analysed. Nine occupants participated during a three-week field study. Thermal preferences were learned by a framework grounded on fuzzy logic used to calculate a set point implemented in the HVAC control system of an open-plan office. The results show that only 29% of the occupants’ thermal comfort improved and decreased on 71% of them, when implementing the user-driven control strategy. The performance of the demand-driven control strategy was mainly influenced by insufficient and poorly distributed data, as well as the effect of thermal expectations on occupants’ thermal responses. These were affected by the better opportunities of behavioural adjustment offered by the participatory sensing approach as well as of their experience with the prevailing outdoor environment. The findings indicate that implementing demand-driven control grounded on user feedback should consider constant data renovation, accurate measurements of local indoor climate and occupants’ thermal expectation.
Info
Journal Article, 2019
UN SDG Classification
DK Main Research Area
Science/Technology
