一种用于吸波超材料快速设计的特征辅助优化方法
By constructing a feature-assisted optimization model and combining machine learning and coupled-mode theory, the problem of the neural network coupled-mode method getting trapped in local minima in metasurface design is solved, and efficient optimization and rapid electromagnetic response prediction of metasurface design are achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- SOUTHEAST UNIV
- Filing Date
- 2024-03-08
- Publication Date
- 2026-07-17
AI Technical Summary
Existing neural network coupled-mode methods are prone to getting trapped in local minima in metasurface design, resulting in low optimization efficiency, and traditional full-wave electromagnetic simulation calculations consume a lot of time and memory.
By combining machine learning and coupled-mode theory, a feature-assisted optimization model is constructed through grid sampling, full-wave electromagnetic simulation, coupled-mode parameter database, and multilayer perceptron neural network. The resonant frequency is used as the feature parameter to formulate the optimization objective function, avoid local optima, and quickly correct the absorption band.
It reduces the time of metasurface design optimization by two orders of magnitude, improves the ability to avoid local optima, and enhances design efficiency and accuracy.
Smart Images
Figure CN118072887B_ABST