Secure Aggregation-Based Big Data Analysis and Power Prediction Model for Photovoltaic Systems: A Multi-Layered Approach

This study presents a novel approach to enhancing the security and accuracy of photovoltaic (PV) power generation predictions through secure aggregation techniques. The research focuses on key stages of the PV data lifecycle, including data collection, transmission, storage, and analysis. To safeguard against potential attacks and prevent data leakage across these critical processes, Paillier and Brakerski–Gentry–Vaikuntanathan (BGV) homomorphic encryption methods are employed. By integrating the transport layer security (TLS) protocol with edge computing during data transmission, this study not only bolsters data security but also minimizes latency and mitigates threats. Robust strategies for key management, access control, and auditing are implemented to ensure monitored and restricted access, further enhancing system security. In the analysis phase, advanced models such as Long Short-Term Memory (LSTM) networks and Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (CEEMDAN) are utilized for precise time-series predictions of PV power output. The findings demonstrate the effectiveness of these methods in managing large-scale PV datasets while maintaining high prediction accuracy and strong security measures.