Enhancing building energy efficiency using a random forest model: A hybrid prediction approach

The building envelope considerably influences energy consumption. To enhance the efficiency of buildings, this paper proposes an approach to predict consumption based on design envelope. parameters include comprehensive heat transfer coefficient and solar radiation absorption exterior walls, roof, outer windows, window-wall ratio. is applied optimize structure a university teaching in northern China. First, information model established Revit imported into DesignBuilder analysis software. Subsequently, data set abovementioned 6 obtained by performing orthogonal testing simulations. On basis, RF used rank importance each parameter, Pearson function evaluate corresponding correlations. results show that most important with highest correlations are coefficients walls windows Finally, prediction compared BP artificial neural network (BP-ANN) support vector machine (SVM). findings indicate exhibits notable advantages optimal among models.