DETERMINATION OF BUILDING HEATING AND COOLING ENERGY NEEDS USING ARTIFICIAL NEURAL NETWORK

This paper presents the results of Artificial Neural Network (ANN) based study, which will be used in prediction of building energy requirement. Building heating and cooling requirements were estimated by benefitting building form factor, orientation angle, insulation thickness and transparency ratio parameters as inputs. Because of the successful results achieved, a backpropagation ANN structure with Levenberg Marquardt training function is preferred and data presented to network by being normalized. The data used for training and testing of ANN were calculated by using explicit finite difference method for a brick wall. The numerical applications were carried out in FORTRAN program by using real average climatic data of Elazig region. ANN applications were done with MATLAB program. When the outputs obtained from ANN were compared with numeric results, it is seen that the ANN based application predicts the building energy requirements with the accuracy ratio varying between 93–99%.