jismart2024071: Deep Learning-Based Assessment of Rooftop Photovoltaic Potential in Rural Areas: A Case Study of Qingdao, China

Abstract

The depletion of non-renewable energy resources and their environmental impact highlight the urgent need for sustainable energy solutions. This study focuses on leveraging the untapped solar energy potential of rural rooftops, proposing an automated approach to assess and optimize photovoltaic (PV) installations in Qingdao, China.Using high-resolution remote sensing imagery and advanced deep learning techniques, including U-Net and SegFormer, the research constructs a comprehensive dataset from Google Earth imagery. Solar radiation models are integrated to estimate PV energy output, accounting for roof slope, orientation, and ground reflectivity. The results reveal significant solar energy potential in Qingdao’s rural areas, with high accuracy in identifying suitable rooftops. Additionally, the study evaluates the economic benefits of PV adoption, demonstrating potential reductions in electricity costs for rural households.This research provides a novel, data-driven framework for large-scale solar potential estimation, offering practical insights for integrating solar power into rural infrastructure planning and advancing sustainable energy development.