Artificial neural network models using thermal sensations and occupants' behavior for predicting thermal comfort

It is important to create comfortable indoor environments for building occupants. This study developed artificial neural network (ANN) models for predicting thermal comfort in indoor environments by using thermal sensations and occupants' behavior. The models were trained by data on air temperature, relative humidity, clothing insulation, metabolic rate, thermal sensations, and occupants' behavior collected in ten offices and ten houses/apartments. The models were able to predict similar acceptable air temperature ranges in offices, from 20.6 degrees C (69 degrees F) to 25 degrees C (77 degrees F) in winter and from 20.6 degrees C (69 degrees F) to 25.6 degrees C (78 degrees F) in summer. The occupants' behavior in multi-occupant offices was more complex, which would lead to a slightly different prediction of thermal comfort. Since the occupants of the houses/apartments were responsible for paying their energy bills, the comfortable air temperature in these residences was 1.7 degrees C (3.0 degrees F) lower than that in the offices in winter, and 1.7 degrees C (3.0 degrees F) higher in summer. The comfort zone obtained by the ANN model using thermal sensations in the ten offices was narrower than the comfort zone in ASHRAE Standard 55, but that obtained by the ANN model using behaviors was wider than the ASHRAE comfort zone. This investigation demonstrates alternative approaches to the prediction of thermal comfort. (C) 2018 Elsevier B.V. All rights reserved.