基于PSO的级联马赫-曾德尔型光滤波器优化设计

Optimal Design for Cascaded Mach-Zehnder Interferometer Based Filter Using PSO

首次应用连续和离散版本的粒子群优化算法(PSO),实现级联马赫-曾德尔干涉仪(MZI)型光滤波器中关键结构参数(耦合角和相移因子)对理想系数的逼近,完成光滤波器的优化设计.结果表明:三级级联MZI型滤波器输出响应与理想响应之间的误差控制在10-5量级内;同时与遗传算法(GA)设计方案相比,PSO方案具备计算时间少、逼近适应度好、滤波消光比高的特点.

Both continuous and discrete particle swarm optimization （PSO） algorithms are applied to design cascaded MachZehnder interferometer （MZI） based optical filter, for the first time. Then the optimal structural parameters （i. e. coupling angles and phase-shift factors） are obtained,by approaching the ideal coefficients with coupling angles and phase-shift factors.In our experiment, a filter of 3-cascaded MZI has been shown, and the relative error between the obtained response and ideal response is about 10^-5. Meanwhile, compared with genetic algorithm （GA） scheme, the PSO scheme is characterized by short computing time, excellent approach fitness and high extinction ratio.