Application of a neuro-fuzzy model to evaluate the thermal performance of typical Australian residential masonry buildings

This paper describes the utilisation of a predictive model for studying the thermal performance of masonry housing based on a neuro-fuzzy approach using a set of training data collected from four test house modules. The room temperatures for the four modules have been predicted from a collection of interior surface temperatures using the ANFIS platform. It has been shown that for the test modules the ANFIS Sugeno-type modelling approach offers an accurate and reliable prediction tool by which given input-output patterns could be achieved with a satisfactory level of accuracy.