Intelligent soft-computing based modelling of naturally ventilated buildings

Abstract

The paper presents recent results on the application of the soft computing methodology for modelling of the internal climate in office buildings. More specifically, a part of a recently completed naturally ventilated building is considered which comprises three neighbouring offices and one corridor within the Portland Building at the University of Portsmouth. The approach adopted uses fuzzy logic for modelling, neural networks for adaptation and genetic algorithms for optimisation of the fuzzy model. The fuzzy models are of the Takagi-Sugeno type and are built by subtractive clustering. As a result of the latter, the initial values of the antecedent non-linear membership functions and the consequent linear algebraic equations parameters are determined. A method of extensive search of fuzzy model structures is presented which fully explores the dynamics of the plant. The model parameters are further adjusted by a back-propagation training neural network and a real-valued genetic algorithm in order to obtain a better fit to the measured data. Results with real data are presented for two types of models, namely Regression Delay and Proportional Difference. These models are applied for predicting internal air temperatures.