Federated optimal control of interconnected smart grids

Abstract

This paper proposes a federated control framework for the optimal regulation of interconnected smart grids. Unlike conventional centralized strategies, which require extensive data sharing from all nodes, our method preserves privacy by letting each local subsystem solve its own Riccati equation, generating control signals independently while sharing only minimal updates with a central aggregator. The aggregator then refines these partial solutions using the full system model, achieving near-optimal performance without disclosing sensitive local measurements. We also develop a hybrid control mechanism that dynamically switches trust factors for nodes under cyber attack, ensuring stability and resilience even in strongly coupled conditions.

Simulation results demonstrate that our federated approach yields cost performance comparable to a fully centralized LQR design, yet operates with significantly lower communication overhead. Moreover, the hybrid controller effectively mitigates cyber intrusions, preventing single-node compromises from destabilizing the entire grid. Hence, the proposed scheme addresses pressing challenges in modern smart grids – namely data privacy, cyber resilience, and scalability – while retaining the benefits of global coordination. In addition to linearized dynamics, future extensions will focus on nonlinear systems and advanced privacy-preserving protocols, paving the way for robust, next-generation energy networks.