Fuzzy pattern recognition on thermal comfort in cold area

Because human thermal sensation is always defined subjectively, we could not point out the difference between "hot" and "warm" property, so fuzzy sets theory is used here to set up a system of fuzzy pattern recognition to evaluate indoor thermal environment. The results of a field study of thermal environment and thermal comfort in residential buildings in Harbin are fuzzy statistical analyzed. The subordinate functions to different levels of thermal sensation are determined. The characteristic variables are air temperature, radiation temperature and thermal insulation of clothing, which are selected with genetic algorithm of elitist selection and steady state. The significant coefficients of the optimized characteristic variables above are determined. A model of fuzzy comprehensive evaluation to thermal sensation is established. The optimum temperature range is 18similar to24degreesC in winter according to the principle of maximum subordinate to the levels of different thermal sensation.