With the increasing global energy demand and the severity of environmental issues, photo- voltaic power generation has received increasing attention as a clean and renewable energy source. However, the unique environment and complex nature of photovoltaic power plants have posed challenges for fault diagnosis and operational optimization. To address this issue, intelligent operation and maintenance technology for island photovoltaic systems has emerged. This technology utilizes artificial intelligence, the Internet of Things, and big data to achieve real-time monitoring, fault diagnosis, and operational optimization of photovoltaic power plants, thereby improving their efficiency and reliability. Among these technologies, artificial intelligence plays a central role. In this paper, we propose a fault analysis and pre- diction model based on multimodal data, which employs customized multimodal learning methods to enhance the accuracy of fault analysis in photovoltaic power plants.