Supercontinuum Generation in Suspended Core Fibers Based on Intelligent Algorithms

In this study, a suspended-core fiber structure, combined with neural networks and optimization algorithms, is utilized to design a supercontinuum (SC) generation method that balances the target spectral range with optimal flatness. The design also enables the reverse calculation of key physical parameters of the fiber structure and the initial pulse. Under a -20 dB intensity calibration, the spectral coverage can reach up to 16 μm. Additionally, the proposed method can reverse-engineer structural parameters such as the fiber core diameter, suspended width, and fiber length, as well as physical parameters such as the peak power, full width at half maximum (FWHM), and chirp of the initial pulse to meet specific supercontinuum targets. The supercontinuum spectral bandwidth achieved in this study is among the highest reported in similar research. Furthermore, the initial pulse required to generate a supercontinuum with a spectral range exceeding 20 μm requires a lower peak power compared to the power levels required in previous studies to achieve the same coverage. By integrating optimization algorithms with a neural network model, the proposed intelligent algorithm significantly reduces computational costs. The reliability of the fiber structure is validated through an analysis of spectral coherence and manufacturing tolerance performance, providing an important theoretical foundation for supercontinuum design and fiber parameter optimization.