The rapid growth of urban populations has intensified the global demand for clean and renewable energy sources. Among these, solar power has emerged as a vital element in sustainable urban planning. Integrating solar photovoltaic (PV) systems into city energy infrastructures represents a key step toward achieving sustainable urban transitions. Nevertheless, fluctuating weather conditions create significant challenges for optimizing solar energy generation and its seamless integration into urban energy networks. This study introduces an artificial intelligence-driven predictive modeling framework designed to support the development of sustainable urban solar energy systems. The proposed approach utilizes advanced machine learning algorithms to predict the degradation of solar energy systems by incorporating meteorological variables and urban air quality indicators within the densely populated capital of Bangladesh. The model is trained and validated using historical weather data alongside real-time degradation records from solar installations in Dhaka, Bangladesh. The results indicate that the proposed predictive model notably improves forecasting accuracy. This research highlights the potential of machine learning as a robust and precise tool for modeling complex urban solar energy dynamics. Furthermore, the developed framework offers practical value to urban planners, utility managers, and system operators by enhancing operational performance, supporting grid integration, and improving the financial sustainability of solar projects. The insights gained contribute to advancing smarter, more resilient, and sustainable urban energy infrastructures.