摘要:
随机并行梯度下降(SPGD)算法是实现大功率主振荡功率放大器相干合成的有效手段之一.分析了SPGD算法用于相干合成的基本原理,设计了基于SPGD算法的相位控制系统,给出了基于该SPGD控制器的相干合成实验结果.理论分析表明,SPGD算法控制器的迭代速率为200 kHz,对于两路相干合成的平均控制带宽大于12.5 kHz,最优控制精度可达1/179个波长.实验表明,SPGD控制器能够实现高速的相位控制.在相位噪声幅度大于1/10个波长、频率为3 kHz的情况下,有效地实现了两路激光的相位锁定,相位残差控制在1/25个波长内.
Abstract:
Stochastic parallel gradient descent (SPGD) algorithm is one of the effective ways to realize high power fiber laser coherent combining of master oscillator power amplifier. The basic principle of the SPGD algorithm in the coherent combining is introduced, the SPGD phase controller has been designed and manufactured, and the representative experiment results are presented. Theoretical analysis shows that the iteration rate is 200000 per second, the average compensation bandwidth is about 12500 Hz, the minimal phase control error is 1/179 wavelength. Experiment shows that the SPGD controller can perform well in phase locking of two lasers, the average phase residual error is less than 1/25 wavelength when the amplitude of the phase noise is greater than λ/10, and the frequency is 3 kHz.