AI-Driven Green Campus: Solar Panel Fault Detection Using ResNet-50 for Solar-Hydrogen System in Universities

The solar-hydrogen system installed at the Samcheok Campus of Kangwon National University is the setting for this paper's presentation of an AI-driven method for fault detection in solar panels. Ensuring the best performance and reliability of renewable energy systems is becoming increasingly important in educational institutions, such as universities. We suggest using the deep learning architecture ResNet-50 to automatically identify solar panel defects. We improved the ResNet-50 model for fault classification by preprocessing the dataset obtained from the campus's solar-hydrogen system. We evaluate the proposed method using metrics such as data accuracy and loss, training and validation loss, and training and validation accuracy, and we compare its performance with state-of-the-art defect identification systems. The experimental findings highlight the AI-driven defect detection system's effectiveness and its potential for in-the-moment solar panel installation monitoring and maintenance. We demonstrate the usefulness and advantages of integrating AI technology into green campuses using the case study of Kangwon National University Samcheok Campus. This work establishes the foundation for future research in AI -driven fault detection systems for renewable energy systems and advances sustainable energy management practices in educational institutions.