Classification algorithm-assisted multi-objective optimization in terahertz hollow core anti-resonant fiber design

This study applies multi-objective optimization algorithms to the design of hollow anti-resonant fiber structures in the terahertz frequency band, focusing on four distinct models. It introduces a classification algorithm as a surrogate model to assist in multi-objective optimization, addressing challenges related to computational efficiency and data classification. The optimization targets key performance indicators, including low loss, high higher-order mode extinction ratio, high birefringence, and high polarization extinction ratio. Multiple multi-objective optimization algorithms are compared to evaluate their performance. The classification algorithm aids in parameter tuning and facilitates a comparative analysis of different optimization algorithms, enhancing the overall optimization process.