Data center cloud edge collaborative energy management method and system based on reinforcement learning

By combining cloud-edge collaborative architecture with reinforcement learning and model predictive control, the problem of system model mismatch in data center energy management is solved, and an efficient and economical energy management solution is achieved.

CN122315919APending Publication Date: 2026-06-30SHANDONG UNIV
View PDF 0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN Β· China
Patent Type
Applications(China)
Current Assignee / Owner
SHANDONG UNIV
Filing Date
2026-05-29
Publication Date
2026-06-30

AI Technical Summary

Technical Problem

In data center energy management, existing technologies show that reinforcement learning system models have poor adaptability and are unable to cope with model mismatch problems caused by changes in system operating conditions and parameter mutations.

Method used

The cloud-edge collaborative architecture is adopted. The cloud side uses reinforcement learning to build the system model and dynamically updates it by combining the real-time status feedback from the edge side. The edge side uses model predictive control for rolling optimization and realizes closed-loop self-updating of the system model through real-time deviation feedback.

Benefits of technology

It improves the adaptability and accuracy of the system model, responds in real time to fluctuations in new energy power generation and load changes, reduces overall operating costs, and enhances the system's robustness and adaptability under complex operating conditions.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122315919A_ABST
    Figure CN122315919A_ABST
Patent Text Reader

Abstract

This invention discloses a data center cloud-edge collaborative energy management method and system based on reinforcement learning, relating to the fields of data center energy management and cloud computing technology. Cloud-side execution: A system model is constructed based on historical system operation data; real-time operating status feedback from the edge side is received; and the parameters of the system model are dynamically updated according to the real-time operating status. Furthermore, the power generation capacity of new energy sources and the power demand of the load are predicted, generating prediction data, and the updated system model and prediction data are distributed to the edge side. Edge-side execution: The system model and prediction data are received; model predictive control is adopted; under the condition of satisfying system constraints, rolling optimization is performed with the goal of minimizing the overall operating cost, solving for the control sequence and executing it; the real-time operating status is fed back to the cloud side, enabling the cloud side to correct and update the parameters of the system model, achieving closed-loop self-updating of the system model. This solves the problem of energy management strategy failure caused by system model mismatch.
Need to check novelty before this filing date? Find Prior Art