Analyzing the energy performance of buildings by neuro-fuzzy logic based on different factors

Energy performance of buildings is an important issue to estimate the energy waste of buildings and their impact on the environment, so designing energy-efficient buildings could improve their energy performance. In this case, the estimation of heating and cooling loads plays an important role in this regard. However, there are few factors with unpredictable influences on the heating and cooling loads. This study has attempted to analyze the eight parameters that can significantly affect the heating and cooling loads through the Neuro-fuzzy logic approach. Accordingly, the eight parameters of relative compactness, surface area, wall area, roof area, overall height, orientation, glazing area and glazing area distribution were considered as inputs and predicting the cooling and heating load changes was regarded as the output of this study. The model was developed and its results were measured in two regression indicators of r and RMSE. Based on the obtained results it was found that roof has the strongest impact on heating and cooling loads (RMSE: 4.3596), moreover, if two factors were concurrently changed, then the combination of relative compactness and wall area can significantly affect the heating and cooling loads (RMSE: 2.6312). The most influential combination of three factors is observed as well for the heating and cooling load and these factors are relative compactness, wall area and glazing area (0.5948). The most influential combination of three factors is observed as well for the hearing load and these factors are relative compactness, wall area and glazing area (RMSE: 0.5948 and RMSE: 1.5769 for heating and cooling load, respectively). However, Neuro-fuzzy logic showed overfitting for more than two inputs, therefore it is not recommended for more than two inputs.