一种基于深度图学习的叶轮机械叶片颤振边界预测方法
By training flow field prediction networks and flutter parameter recognition networks based on deep graph learning, the problems of complexity and computation time in predicting flutter boundaries of turbomachinery blades are solved. This enables rapid and accurate prediction from blade control parameters to flutter characteristic parameters, simplifies the calculation process, and reduces optimization design costs.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- XI AN JIAOTONG UNIV
- Filing Date
- 2022-08-11
- Publication Date
- 2026-07-17
AI Technical Summary
Predicting the flutter boundary of turbomachinery blades is complex. Traditional methods are time-consuming, have poor convergence, and are costly to optimize. Furthermore, it is difficult to quickly and effectively predict the flutter characteristics from blade profile control parameters to flow field parameters.
A deep graph learning-based approach is used to train a flow field prediction network and a flutter parameter recognition network. Through adaptive mesh partitioning and graph structure representation, the rapid prediction of blade geometric parameters, flow field parameters, and flutter characteristic parameters is achieved. A deep graph convolutional neural network is used for direct prediction of aerodynamic damping.
It enables rapid prediction of blade flutter boundaries, improves prediction accuracy and efficiency, reduces dependence on mesh structures, simplifies the calculation process, and reduces optimization design costs.
Smart Images
Figure CN115292844B_ABST