A Knowledge-driven Energy Management System Based on Cloud-edge-end Closed-loop Architecture

With the rapid growth of energy demand and development of distributed energy systems (DESs), the coupling of energy conversion devices and various energy flows within the system is gradually strengthened, and thus the design and application of energy management strategy and energy management system (EMS) confront novel challenges. This paper proposes a knowledge-driven energy management strategy and develops an EMS based on cloud-edge-end closed-loop architecture for DESs. The essence of cloud-edge-end closed-loop architecture is to integrate the powerful computational resources of cloud computing with the rapid response capability of edge computing, while introducing a closed-loop feedback channel to collect real-time system operation and energy supply-demand information on end side. This integration facilitates the dynamic knowledge strategy updating of cloud side and model updating edge side. The knowledge strategy is employed to calibrate the parameters and constraints of the algorithm in accordance with the system real-time state. The case study shows that the application of the EMS achieves a reduction in the unit load cost and unit load carbon emissions by an average of 8.68% and 7.06%, respectively. Furthermore, the knowledge-driven energy management strategy significantly accelerates the convergence of the algorithm, with a solution time reduction of 40.21% relative to the particle swarm optimization (PSO) algorithm. This enhancement provides a new approach to achieve more efficient, accurate and rapid energy management.