基于随机并行梯度下降算法的湍流像差校正仿真  

Simulation of turbulence aberration correction based on stochastic parallel gradient descent algorithm

作　　者：马慧敏[1] 张京会[1] 张鹏飞[1] 范承玉[1] 

机构地区：[1]中国科学院安徽光学精密机械研究所中国科学院大气成分与光学重点实验室,安徽合肥230031

出　　处：《红外与激光工程》2011年第9期1738-1742,共5页Infrared and Laser Engineering

基　　金：国家自然科学基金(10574127);中国科学院合肥物质科学研究院计算中心资助课题

摘　　要：随机并行梯度下降(SPGD)算法可不依赖波前探测直接优化系统性能指标来校正畸变波前。建立了基于随机并行梯度下降算法控制的61单元湍流校正仿真模型,实现了通过该算法控制倾斜镜和变形镜对湍流引起的像差的校正。结果发现,该算法能够找到补偿湍流像差所需的倾斜镜和变形镜的最优面形。采用SPGD算法控制,倾斜镜校正后,远场光斑质心更靠近轴心而且轴上斯特列尔比有所提高;倾斜镜和变形镜共同校正比变形镜单独校正的效果好,这也说明倾斜镜的校正是有效的。Being independent of wavefront detecting, the stochastic parallel gradient descent （SPGD） algorithm can optimize the system performance metric directly to correct wavefront distortions. A 61- elements turbulence correction model based on SPGD algorithm was simulated, and distortions induced by atmosphere turbulence were corrected through controlling both tilt mirror and deformable mirror by the algorithm. Simulation results demonstrate that the SPGD algorithm can find the optimum controlling shapes of tilt mirror and deformable mirror which are applied to correct turbulence aberration. After being corrected by tilt mirror controlled by SPGD algorithm, the facular centroid is closer to the axis center and the strehl ratio of axis is increased in the far-field; the results of the effect corrected by tilt mirror and deformable mirror are better than that only by deformable mirror, which illustrates that the correction based on SPGD algorithm by tilt mirror is effective.

关 键 词：自适应光学 随机并行梯度下降算法 湍流 

分 类 号：TP273.2[自动化与计算机技术—检测技术与自动化装置]