改进遗传单纯形混合算法在光纤对接中的应用

本文首先给出了纤端光场分布的3D模型并针对单模光纤与激光器的对接分析了错位、角度偏差等对接效率影响因素。通过与传统的爬山法进行比较,提出将遗传单纯形混合算法应用到光纤对接中。单纯形法是一种局部搜索方法,这种寻优方法收敛速度快,它与遗传算法组合起来可以改善单独使用遗传算法收敛速度慢的缺陷。同时,针对光纤对接要求初始种群小所造成的个体多样性差的缺点,改进了单纯形的反射、压缩系数,有效地避免了遗传操作的过早收敛,并通过仿真结果证明了混合算法可以在10代之内找到光纤对接的最佳耦合点。

Application of improved simplex genetic algorithm hybrid in automatic optical fiber alignment

This paper first gives a 3-D profile of light intensity in fiber end-face and analyzes the consequences of initial offset and angular misalignment between a single-mode fiber and a laser diode coupling. Compared with conventional Hill-climbing algorithm, a novel method for optical fiber alignment using Simplex Genetic Algorithm Hybrid (SGAH) is presented. The Simplex method is a local searching method. Combined with Genetic Algorithm (GA), this method converges quickly and can amend the defect that GA converges slowly. Considering the poor individual diversity caused by little population of fiber alignment, the expansion and contraction coefficients of simplex method are improved. Simulation result also shows that new algorithm can reach the best coupling point within 10 generations.