A novel sensitivity analysis of residential building retrofitting strategies: Artificial neural network vs. linear regression

Abstract

This study compares the implementation of traditional methods (linear regression) and artificial neural network (ANN) algorithms in sensitivity analysis for residential building retrofitting strategies. The impact of building retrofitting strategies, such as wall insulation, roof insulation, glazing types, and shading systems, on the life-cycle carbon emissions (LCCO2) and life-cycle costs (LCC) of a residential building in the severe cold climate of China is investigated. The results demonstrated that the ANN modelling was more accurate and stable than the linear regression method. Through impact factor analysis, it was found that window area and insulation types strongly impact the LCC, while shading, insulation type and insulation thickness strongly influence the LCCO2 in the building. This study validated the feasibility and efficiency of the ANN methodology in sensitivity analysis for residential building retrofitting strategies.