Study on sensor fusion for detecting human´s thermal comfort considering of individuals

This paper presents the methodology how to utilize sensor networks in order to predict human´s thermal comfort and sensation. The neural network was dynamically organized on the basis of correlations with the thermal sensation of the occupants and many other values in the sensor network, and the structure of the neural network was updated cyclically. In this paper, the air-conditioning system in an office is used as an example. We place 21 temperature sensors and subjects in the indoor environment, and conduct an experiment where the sensor readings and the thermal sensation of the subjects are associated. Finally, the experimental results are shown, where unspecified sensors are chosen properly and used for predicting the desired value.