Bayesian networks for thermal comfort analysis

To analyze the causal relationships between thermal comfort and the related factors, this paper presents a model of Bayesian networks and its experimental simulation. The samples used in the experimentation consisted of 1,584 (22 participants under 9 conditions tested 8 times) pieces of questionnaire data which included ratings by the 22 participants of their thermal sensation votes in five bodily regions (face, chest, hands, buttocks, and soles of the feet). The proposed Bayesian network consists of four layers: the external environment, thermal sensation change, thermal sensation votes and comfortability. Assigning evidence to the thermal comfort for the Bayesian network, the experimental simulation shows that the thermal sensations of the face play an important role in a person´s overall comfortability. The sensitivity of the model is also discussed.