The Study Of Stochastic-parallel-gradient-descent Adaptive Optics Control Platform Based On Dsp

Stochastic parallel gradient descent (SPGD) algorithm is a gradient descent algorithm, which measures gradient estimation by perturbing all variables simultaneously. Compared with the conventional optimization algorithms, the SPGD algorithm has the advantages of fast convergence speed and high convergence precision. Recently, the SPGD algorithm has been widely used in wavefront distortion correction and becomes a research focus in adaptive optics field. To meet the requirement of real-time control and the flexibility of the SPGD algorithm, the DSP is used as the SPGD controller in the adaptive optics system. Firstly, the SPGD adaptive optics control platform based on DSP is designed and set up. The design of the hardware and the software, the working principles of every function module and the data transfer are discussed in detail. Secondly, the SPGD adaptive optics system based on DSP is set up after the conventional 37-uint adaptive optics system is rebuilt. Then the closed-loop experiments are implemented and the wavefront corrections of the static aberration introduced by the deformable mirror and the dynamic artificial turbulence are realized. Results of the experiments indicate that the application of DSP in the SPGD adaptive optics system is feasible and the SPGD algorithm is robust and convergent. The thesis has laid a foundation for the study of the practicality of the wavefront control technique based on the SPGD algorithm.