A Secure Federated Load Forecasting Scheme for Power Distribution in Smart Grids

In this paper, we present a novel approach for service-oriented and smart grid load forecasting, designed to address data silo issues and enable secure communication. To tackle the challenges, we propose a hybrid cryptographic algorithm based on federated learning and, the SM series algorithms tailored for secure data transmission. Specifically, we implement an improved SM9 algorithm that combines homomorphic properties with improved efficiency. Additionally, we refine and integrate SM2, SM3, and SM4 algorithms to counteract identity spoofing, poisoning attacks, and other network threats. Experimental results demonstrate that, compared to existing schemes, our approach achieves higher accuracy and improved stability, while also offering enhanced security in real-world applications and ensuring the safe forecast of power system load.