Method, system, device and medium for multi-time scale prediction of data center load
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- STATE GRID ANHUI ELECTRIC POWER CO LTD
- Filing Date
- 2026-04-21
- Publication Date
- 2026-07-17
AI Technical Summary
Existing technologies cannot adapt to the multi-timescale non-stationary characteristics of data center loads, resulting in insufficient load forecasting accuracy, lack of adaptability in error correction, and unreasonable backup power decisions.
The optimal number of modes and penalty factor are adaptively determined by Bayesian information criterion. The frequency characteristics of the protection load and computing load are separated by variational mode decomposition. A long short-term memory neural network model is constructed for prediction. The weighted integration is carried out through state-aware attention mechanism. The error is corrected by conditional Gaussian mixture error model. Finally, the backup power capacity is dynamically determined by calculating the conditional excess expected loss in a two-step method.
It improves load forecasting accuracy, achieves adaptive dynamic balance between error correction and the economy and reliability of backup power decision-making, and solves the problems of decreased forecasting accuracy and unreasonable backup power decision-making in existing technologies.
Smart Images
Figure CN122415266A_ABST