Intelligent modelling of the indoor climate in buildings by soft computing.

摘要: The paper considers the application of soft computing techniques for predictive modelling in built sector. TakagiSugeno fuzzy models are by subtractive clustering to provide initial values antecedent non-linear membership functions parameters and consequent linear algebraic equations coefficients. A method extensive searching possible model structures is presented which explores all permutations a specified range orders derive model. further adjusted back-propagation neural network real-valued genetic algorithm order obtain better fit measured data.