摘要
A hybrid optimal algorithm, named the SAA-PA in brief, based on the simulated annealing algorithm （SAA） and the Powell algorithm （PA） is proposed. The proposed algorithm puts the random search strategy of the SAA into the PA, which can prevent optimizing courses from trapping in local optima. The SAA-PA can effectively solve multimodal optimization in the distributed multi-pump Raman amplifier （DMRA）. Optimal results show that, under the conditions of the on-off gain of 10 dB, the gain bandwidth of larger than 80 nm and the fiber length of 80 km, the gain ripple of less than 1.25 dB can be designed from the DMRA with only four backward pumps after the optimization of the proposed SAA-PA. Compared with the pure SAA, the SAA-PA can attain a lower gain ripple with the same number of pumps. Also, the relationship between the optimal signal bandwidth and the number of pumps can be simulated numerically with the SAA-PA.
基于模拟退火算法(SAA)和鲍威尔算法(PA)提出了一种新的混合优化算法——模拟退火鲍威尔算法(SAA-PA).该算法将模拟退火算法的随机搜索策略纳入到鲍威尔优化算法中,能使优化解不陷入局部最优从而获得全局优化解.该优化方法可以有效地解决多目标的优化问题,特别适用于拥有多个局部最优值的分布式多泵浦拉曼放大器(DMRA)的优化问题.仿真结果显示,在80km传输光纤上只要4个后向泵浦就能实现开关增益10dB,带宽大于80nm,增益平坦度小于1.25dB的平坦增益.与单纯的模拟退火算法的优化结果相比,在相同数目的泵浦条件下所得优化结果的增益谱特性有了显著的提高.同时,该方法可以较方便地仿真出信号增益带宽与泵浦数目的内在关系.
基金
The Start-Up Research Foundation of Nanjing Uni-versity of Information Science and Technology (No.QD60)
