Power transmission line defect detection method based on privacy protection of hierarchical federated learning
By employing a dynamic client-side grouping method based on hierarchical federated learning and two-factor clustering, the problem of data silos is solved, enabling high-precision and highly generalized transmission line defect detection, thus meeting the needs of power inspection departments with limited computing resources.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- INFORMATION & COMM BRANCH OF STATE GRID INNER MONGOLIA EAST ELECTRIC POWER CO LTD
- Filing Date
- 2026-01-27
- Publication Date
- 2026-06-09
AI Technical Summary
Existing technologies are insufficient to effectively aggregate heterogeneous data resources scattered across different regions and train high-precision, highly generalizable transmission line defect detection models while ensuring data privacy and security. Furthermore, power inspection departments with limited computing resources struggle to support large-scale collaborative model training.
By adopting a hierarchical federated learning architecture, local transmission line defect data from multiple clients are acquired, and collaborative training is performed using the hierarchical federated learning architecture. Combined with a dynamic client grouping method based on two-factor clustering, the computational and communication load is reduced, enabling collaborative mining and utilization of cross-regional data value.
While protecting data privacy, it integrates scattered data resources, improves the accuracy and generalization ability of the model, adapts to environments with limited computing resources, and achieves efficient and accurate detection of transmission line defects.
Smart Images
Figure CN122176478A_ABST