A Cloud Energy Storage Based Energy Management Framework for Energy Internet

The Energy Internet paradigm relies on intelligent routing to enable flexible and resilient power networks. While existing power routing protocols have demonstrated effectiveness in minimizing losses and managing congestion, energy storage is still commonly treated as a local and isolated resource. This limits the ability of the network to exploit storage as a system-level flexibility service. This paper proposes a cloud energy storage service integrated into an Energy Internet formed by multiple energy routers. Distributed storage elements embedded within selected routers are virtualized into a cloud layer that provides coordinated charging and discharging services in response to energy surplus and demand requests. A novel energy management algorithm is introduced, in which direct source-to-load power transfers are prioritized to reduce losses and storage cycling, while the cloud storage service balances residual power mismatches. The algorithm operates in a distributed manner, explicitly accounting for DC network losses, router efficiencies, and link capacity constraints. The proposed energy management scheme generates packetized power transmission assignments that serve later as inputs to distributed Q-learning-based power routing protocol, enabling adaptive and optimal routing without centralized control.