基于模拟退火遗传算法的微波光子滤波器特性研究  

Characteristics of Microwave Photonic Filter by Using Simulated Annealing Genetic Algorithm

作　　者：蒋宏彬[1,2,3] 刘广智[2,3] 曾小平[1] 杨凡[1] 

机构地区：[1]绵阳师范学院物理与电子工程学院,四川绵阳621000 [2]区域光通信网络系统与新型光电子器件国家重点实验室,北京100871 [3]北京大学信息科学技术学院,北京100871

出　　处：《中国激光》2012年第8期227-231,共5页Chinese Journal of Lasers

基　　金：国家创新科学基金(09C26215102319)资助课题

摘　　要：采用模拟退火遗传算法,分析了非等间距低通滤波器不同抽头下全正对称系数及含负系数的微波光子滤波器的频率响应特性,并与窗函数法进行了对比分析。仿真结果表明:全正对称系数条件下,带抽头数量越大,滤波器的优化性能越好。正负系数滤波器的对比表明,含负系数的滤波器优化性能明显改善,特别是对低通滤波器而言,系数的全正性极大地限制了光学传递函数的取值范围,影响了滤波器的性能。在同一截止频率下,模拟退火遗传算法优化性能指标略好于窗函数法。如果限制通带和阻带的截止频率,即放宽对过渡带的要求,模拟退火遗传算法性能优势更加明显。By means of simulated annealing genetic algorithm, frequency-response characteristics of nonuniform- delay-time microwave photon filter with full/non positive coefficients are analyzed. And the window-function method is applied at the same time. Simulation results show that the more taps lead to better performance at full-positive symmetric coefficient filter. In the case of low-pass filter, full-positive coefficients limit largely factor on the span of transfer function, the characteristics improvement of non full-positive coefficients filter is obvious. At the same cut- off frequency, the optimal performance of simulated annealing genetic algorithm is slightly better than that of the window-function method. Once limiting the cut-off frequency of pass-band and stop-band, and relaxing the requirements of the transition zone, the advantages of simulated annealing genetic algorithm are much more obvious.

关 键 词：光学器件 模拟退火遗传算法 多抽头 均方根误差最小准则 非等间距微波光子滤波器 Q参数 

分 类 号：TN929.91[电子电信—通信与信息系统] TN24[电子电信—信息与通信工程]