A method for urban building type identification and energy saving potential evaluation

With the rapid acceleration of urbanization and the continuous growth of the urban population, urban energy consumption is projected to rise even more. Quantitative analysis of urban building energy is not only conducive to the formulation of relevant policies and evaluation standards, but also can promote energy conservation and emission reduction in urban buildings. However, current urban building energy simulation faces challenges in obtaining detailed building data. Especially, the complexity and diversity of building types, along with the large number of buildings, making it difficult to know the situation of each building type in urban areas. This paper combines spatial analysis methods with the XGBoost (eXtreme Gradient Boosting) model, utilizing GIS data such as building footprint, POI (Point of Interest), AOI (Area of Interest), and land use data, to achieve the identification and classification of 86.83% of buildings in Shenzhen, with an identification accuracy rate of 84.17%. Based on this, the study establishes 18 types of typical building prototype models and conducts energy consumption simulation and energy-saving potential assessment for buildings in Shenzhen, providing a scientific basis for energy conservation and carbon reduction in urban buildings. The research results indicate that among four energy-saving measures, high-performance window retrofitting has the most significant energy-saving effect, with an energy-saving rate of 6.2%; the energy-saving effects of improving the energy efficiency of air conditioning systems and replacing energy-saving lighting are equivalent, with energy-saving rates of 4.82% and 4.14%, respectively; whereas green roofs have a relatively smaller effect on building energy conservation, with an energy-saving rate of 1.27%.