A hybrid method for photonic crystal fiber polarization filter based on artificial neural network and genetic algorithms

In this paper, we proposed a design method for photonic crystal fiber polarization filters incorporating the artificial neural network (ANN) and the genetic algorithm (GA) to achieve high-performance polarization filters. The ANN is trained to accurately learn the relationship between the structure parameter and polarization properties of PCFs, and GA is utilized to select the structural parameters with optimal performance. Confinement loss, bandwidth, and resonance wavelength of the polarization filter serve as parameters in the multi-objective optimization process performed by the GA. Numerical results demonstrate that the designed PCF filter has high performance, with a confinement loss of 1995.75 dB/cm for the y polarization core mode at 1.55 μm, and crosstalk and bandwidth of 1720.15 dB and 932 nm, respectively. It indicates that the hybrid design method is a promising and viable approach for the development of optical applications.