Prediction of Automobile Passenger's Skin Temperature Using a Neural Network.

Abstract

The purpose of our study is to develop a method for estimating the facial skin temperature of an automobile passenger. The facial skin temperature is a good index for evaluating the environment. For the estimation of skin temperature, the rate of change in facial skin temperature was predicted from environmental data using a neural network. Then the facial skin temperature was estimated from the rate of change in facial skin temperature and the initial facial skin temperature calculated from the environmental data. Furthermore, the level of thermal sensation was estimated from the predicted facial skin temperature. By use of a neural network, the rate of change in facial skin temperature could be predicted from the environmental data easily and accurately, and the facial skin temperature could be predicted within ±0.6°C error. This is better than the method in which heat balance equations for the body are used. The thermal sensation could be estimated within ±0.8 error on the scale used.