自适应增益的SPGD算法

SPGD algorithm with adaptive gain

SPGD算法是一种应用广泛的无波前探测自适应光学控制算法。传统SPGD算法中增益系数保持某一固定值不变,随着变形镜单元数的增加,这将导致算法收敛速度变慢及陷入局部极值的概率增大。Adam优化器是深度学习常用的一种优化随机梯度下降算法,它具有增益系数自适应性调整的特点。将Adam优化器自适应调整增益系数的优势与SPGD算法结合起来用于自适应光学系统控制。分别以32、61、97、127单元变形镜作为波前校正器件,不同湍流强度的波前像差作为校正对象,建立了无波前探测自适应光学系统模型。结果表明,优化后的算法收敛速度更快,而且陷入局部极值的概率降低,并且随着变形镜单元数的增加与湍流强度的增大,算法的优势更加明显。以上研究结果为基于Adam优化的SPGD算法的实际应用提供了理论基础。

SPGD is a control algorithm widely used in wavefront sensorless adaptive optics(AO) systems. The gain is commonly set to a fixed value in the traditional SPGD algorithm. With the increase of the number of DM elements, which can easily lead to the slow convergence speed of the algorithm and the increase of the probability of falling into the local extreme value. Adam optimizer is an optimized stochastic gradient descent algorithm commonly used in deep learning. It has the advantage of achieving adaptive learning rate. The advantages of Adam optimizer adaptive gain and SPGD algorithm were combined to realize adaptive gain for AO system control. The simulation model of wavefront sensorless AO system was established with 32, 61, 97 and 127 elements DM as wavefront correction devices respectively, wavefront aberrations with different turbulence intensities as correction objects. The results show that the optimized algorithm converges faster than basic SPGD algorithm and the probability of falling into local extremum decreases. As the number of DM elements increases and the turbulence intensity increases, the advantages of the optimized algorithm are more obvious. The above research results provide a theoretical basis for the practical application of the SPGD algorithm based on Adam optimization.