Solar photovoltaic (PV) systems are complex, dynamic systems influenced by a wide range of environmental and operational factors. Accurately modelling their performance and detecting anomalies require integrating diverse data sources and capturing intricate dependencies. In this talk will present a graph-oriented information fusion framework designed to enhance solar PV performance analysis. The framework begins with the analysis of temperature and irradiance data using graph-based techniques, which are then extended to incorporate additional PV parameters such as module temperature, ambient air temperature, global shortwave and longwave radiation, and power output. By constructing temporal graphs, the framework captures both spatial and temporal dependencies, enabling a holistic understanding of system behaviour under varying conditions. This talk will explain the development of a Temporal Graph Neural Network (TGN) model that leverages these fused data sources for performance prediction and anomaly detection. The TGN model integrates Graph Convolutional Networks (GCNs) and Gated Recurrent Units (GRUs) to simultaneously model spatial and temporal dependencies, offering a robust approach to analyse complex PV system dynamics. The talk will highlight the mathematical foundations of the framework, including network analysis, information fusion, and deep learning, and demonstrate its practical applications in solar PV diagnostics. By combining diverse data sources and advanced graph-based techniques, this framework provides a powerful tool for improving the reliability and efficiency of solar energy systems.