A method and system for locating faults in a wind turbine drivetrain
By fusing explicit mechanical connections and implicit data associations through a dual-flow graph neural network model, the problem of fine-grained localization under rigid topology construction and small sample conditions in wind turbine drive chain fault diagnosis is solved, achieving high-precision, adaptive fault localization and continuous learning.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- NANJING UNIV OF INFORMATION SCI & TECH
- Filing Date
- 2026-03-10
- Publication Date
- 2026-05-26
AI Technical Summary
Existing wind turbine drivetrain fault diagnosis methods struggle to achieve precise localization under complex coupled operating conditions. In particular, they fail to balance physical mechanisms and implicit correlations under small sample conditions, and suffer from rigid topology construction, insufficient physical interpretability, and catastrophic forgetting in incremental learning.
A dual-flow graph neural network model is adopted, combining physical flow and latent flow. By introducing regularization constraints based on physical adjacency and an incremental learning mechanism, a fault location method for wind turbine drive chain is constructed. This method integrates explicit mechanical connections with implicit data associations to achieve high-precision fault location and continuous self-adaptation.
High-precision fault location of wind turbine drivetrain was achieved under small sample conditions, maintaining physical consistency and continuity of fault location results, and improving applicability and anti-disturbance performance in unstable industrial sites.
Smart Images

Figure CN121808524B_ABST
Abstract
Citation Information
Patent Citations
Process industrial fault diagnosis method based on bidirectional long-short-term neural network
CN111859798A
CNN-GCN-based double-sensor aero-engine case fault source acoustic emission positioning method
CN115014789A