基于物理场的器件性能预测方法、装置、服务器及介质
By establishing a parametric geometric model and co-training the generation and prediction subnetworks, the problem of insufficient accuracy and reliability of existing surrogate models in microwave device performance prediction is solved, achieving efficient and accurate electromagnetic performance prediction, which is suitable for microwave device design and optimization under small sample conditions.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- TIANJIN POLYTECHNIC UNIV
- Filing Date
- 2026-04-28
- Publication Date
- 2026-07-17
AI Technical Summary
Existing surrogate models have limited ability to express complex nonlinear electromagnetic behavior, poor generalization, and lack of physical constraints, resulting in insufficient accuracy and reliability in predicting microwave device performance. This makes it difficult to meet the urgent need for rapid design and optimization iteration of microwave devices under the trend of high frequency and miniaturization.
A parameterized geometric model of the device is established, physical field prediction features are generated through a generative subnetwork, and a gating weight mechanism is used in conjunction with the prediction subnetwork to achieve decoupling training of geometric parameters and physical fields, thereby generating refined electromagnetic performance prediction results.
It achieves efficient and accurate prediction of the electromagnetic performance of microwave devices under small sample conditions, reduces computational costs, and improves prediction accuracy and reliability, making it suitable for rapid design and optimization iteration under small sample conditions.
Smart Images
Figure CN122242272B_ABST