Neural network predictor for thermal comfort conditions

This paper is investigated thermal comfort conditions having surface and fresh air introduced directly in a room using neural network predictors. The investigation is divided into two steps. First step; experimental room system result is obtained the Building Physics Laboratory of the Department of Building Services, University of Debrecen. The room could be heated by radiator, on the wall, on the floor and on the ceiling heating system methods. During the operation of the heating system a series of measurements are derived, involving 20 students as subjects during education, in order to establish the effects of intermittency on thermal comfort feeling. The second step of the study; according to experimental results, some neural network predictors are used modelling the experimental room. Four types of ANNs are used to compare each other. From the results, it is noted that the proposed Radial Basis Neural Network gives the best results for analyzing thermal comfort conditions.