针对相干光正交频分复用(OFDM)系统中峰值平均功率比(PAPR)高的问题,对粒子群算法(PSO)、蝙蝠算法(BA)和鸟群算法(BSA)等几种群智能算法进行了研究,采用群智能算法优化OFDM符号的子载波相位,达到降低PAPR的目的。同时,通过动态调整认知系数和学习因子,分别对蝙蝠算法和鸟群算法进行了改进。对100Gb/s、二进制正交振幅调制(4QAM)的相干光OFDM系统的仿真实验表明,PSO、BA、BSA三种智能算法都能有效降低系统的PAPR,且改进BSA和改进BA与原始信号相比可使PAPR分别降低约5.11dB、5.48dB,具有更好的抑制效果;采用群智能算法优化后,系统误码率性能也得到提高,且随着光信噪比的增大,性能提高更加明显。 In order to solve the problem of high peak-to-average power ratio (PAPR) in coherent optical orthogonal frequency division multiplexing (OFDM) systems, several swarm intelligences, such as PSO, BA and BSA, Algorithm is studied, the group intelligent algorithm is used to optimize the subcarrier phase of the OFDM symbol to achieve the purpose of reducing the PAPR. At the same time, bat algorithm and flock algorithm are improved respectively by dynamically adjusting the cognitive coefficient and learning factor. The simulation experiments of 100Gb / s, quadrature quadrature amplitude modulation (4QAM) coherent optical OFDM system show that PSO, BA, BSA three kinds of intelligent algorithms can effectively reduce the system PAPR, and improve the BSA and improve the BA and the original signal phase Than the PAPR can be reduced by about 5.11dB, 5.48dB, with better suppression effect; using group intelligence algorithm optimization, the system bit error rate performance has also been improved, and with the optical signal to noise ratio increases, the performance increase more obvious.