【摘要】 为了满足4 m级地基大口径光学望远镜千单元自适应光学系统的波前处理规模、速度以及控制带宽要求，研究了基于GPU的千单元级自适应光学系统高速波前处理器。介绍了自适应光学系统波前处理方法，并详细论述了基于GPU的波前处理架构的实现与优化方法。最后，通过湍流模拟器进行桌面模拟实验，得到961单元自适应光学系统的动态性能。实验结果表明：基于GPU的961单元自适应光学系统在采样频率为1 500 Hz时，系统0 dB残余误差抑制带宽能够达到100 Hz，满足4 m级地基大口径光学望远镜自适应光学系统的应用需求。更多还原

【Abstract】 In order to meet the wavefront processing scale,speed,and control bandwidth requirements of the thousand-element adaptive optics system of the 4-m-level ground-based large-aperture optical telescope,a GPU-based high-speed wavefront processor for the adaptive optics system is examined. The wavefront processing method of the adaptive optics system is introduced,and the realization and optimization methods of the GPU-based wavefront processing architecture are discussed. A testbed simulation experiment is performed using a turbulence simulator,in order to analyze the dynamic performance of the 961-unit adaptive optics system. The experimental results show that when the sampling frequency of the 961-unit adaptive optics system is 1 500 Hz,the 0 dB residual error rejection bandwidth of the system can reach 100 Hz. This rejection bandwidth value meets the application requirements of the 4-m optical telescope adaptive optics system.更多还原