An abnormality detection method and system for power system data center station
By constructing a data augmentation method based on spatiotemporal pattern guidance rules and diffusion models, a high-fidelity and balanced training dataset is generated, which solves the problems of high cost and unbalanced distribution of training data in power system data centers, and improves the accuracy and reliability of anomaly detection.
Patent Information
- Application Number
- CN202610573227.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-04-28
- Publication Date
- 2026-07-17
AI Technical Summary
The power system data middleware suffers from problems such as high training data cost, insufficient fidelity, and unbalanced distribution, which leads to a decline in the performance of anomaly detection models when identifying abnormal behavior, affecting the stability and security of the system.
By constructing a data augmentation method based on spatiotemporal pattern guidance rules and diffusion models, a high-fidelity and balanced training dataset is generated. An adaptive scaling module is used to dynamically adjust the sample ratio, and the model parameters are optimized by combining gradient descent method, thus achieving end-to-end optimization of data augmentation and anomaly detection.
It effectively captures the physical characteristics and spatiotemporal correlation patterns of the power grid environment, improves the accuracy and reliability of anomaly detection, solves the problem of imbalanced training data, and enhances the operational stability and security of the power system data platform.
Smart Images
Figure CN122412913A_ABST