Multi-Dimensional Coupled Evaluation and Prediction Of Solar Energy Utilization Indicators on Building Surfaces: A Case Study on Non-Enriched Area, China

The utilization of solar energy on urban building surfaces in non-enriched regions is overlooked, leading to challenges in promoting Building Integrated Photovoltaics (BIPV) in these areas. In this study, a novel assessment framework was proposed to enable high-precision, multi-dimensional coupled evaluation of solar radiation dynamics on city-scale building surfaces. An XGB machine learning-based approach was developed to assess the effective solar energy utilization potential and the distribution of the total solar energy across various orientations in typical non-enriched areas of China (Chengdu, Chongqing and Guiyang ) under dynamic thresholds. Multi-dimensional feature indices of building clusters were extracted by geometric morphometrics. The use of an Area of Interest (AOI)-based method segmented and identified the cluster and type of urban building. The results demonstrated: (1) The potential for effective solar energy utilization in Chengdu city and Chongqing city show similar patterns, i.e., Roof > South > North > West > East, while that in Guiyang city shows Roof > South > West > North > East; (2) the total solar radiation in winter is significantly less than in summer, with winter in Guiyang city being about 63% of summer, Chengdu city about 38%, and Chongqing city only 30%; the shading rate of typical urban buildings varies with orientation, with roofs ranging between 3.45%-6.98%, and facades between 34.70%-50.71%; (3) The effective utilization potential across different orientations demonstrates a non-linear function of utilization thresholds. In the future, it will be necessary to analyze the settings of dynamic thresholds in actual utilization scenarios, considering both economic viability and the enhancement of utilization potential.