A5104666595- Advanced Fiber Optic Sensors
- Photonic Crystal and Fiber Optics
- Color Science and Applications
- Image Enhancement Techniques
- Advanced Measurement and Detection Methods
Tianjin University of Technology
2023-2024
An automatic iterative optimization and inverse design method for photonic crystal fiber (PCF) structures is proposed based on convolutional adversarial autoencoder (CAAE) forward prediction neural network. This takes the two-dimensional (2D) material refractive index distribution of cross-section as structural parameter space can achieve generation PCF in 2D space. The random CAAE network, which compress matrix into a 36-dimensional hyperparameter with Gaussian distribution, hyperparameters...
A mapping network and the relationships between two-dimensional refractive index distribution corresponding optical properties of photonic crystal fibers have been established in this paper. The can realize fast accurate prediction mode properties, input variable is material fiber cross-section, not fixed specific structural parameter variables, which more flexible. generalization for other with different lattice arrangements investigated, transfer learning method also proposed. This only...