Study On Non-wavefront Sensor Adaptive Optics Technologhy Based On Stochastic Parallel Gradient Descent Algorithm

Due to the atmospheric turbulence and the changeable motion state, the wave front is strongly distorted and the phase is severely degenerated, which seriously affect the quality of astronomical objects imaging. The conventional adaptive optics techniques is an efficient way to correct the atmospheric turbulence. However the wave-front aberrations brought by the strong scintillation during the transmission path and the fluctuation arisen from receiving the amplitude at the receive plane especially observing astronomical objects at low elevation or dark targets, are the main causes of the conventional adaptive optics(AO) systems detective error.Compared with the conventional AO systems, the adaptive optics without a wave-front sensor doesn’t require wave-front detecting, reforming, it considers the control signal of wave-front corrector as parameters of optimization, the image sharpness function, sterol ratio, and the received light energy as object function,optimizes the object function in iteration method to achieve the ideal correction effect.There are some algorithms have been commonly used in the control algorithm of the adaptive system at present, stochastic parallel gradient descent(SPGD)algorithm is a control algorithms with multiple parameters used in model free systems, with the advantages of simple parameters, high convergence speed, and good stability, etc. In many applications, SPGD algorithm is becoming the first option while solving the optimizing problems which can’t be solved by analytic method.In the view of SPGD algorithm, the main contents of this paper are as follows:In this paper, a method to optimize the SPGD algorithm by using the power spectrum method is proposed. The power spectrum method can be used to produce the phase perturbation of atmospheric turbulences which accords with Von Karman spectrum and the random disturbance voltage in the algorithm to improve the statistical correlation between the wave-front aberration to be corrected and the phase perturbation caused by the random disturbance voltage, so as to achieve the purpose to increase the convergence speed for the traditional SPGD algorithm. a model of SPGD adaptive optics system is built up based on Matlab, the results show that the method proposed in this paper can achieved a better correction effect and a higher convergence speed.SPGD algorithm software controller is designed and implemented in FPGA and each module has been designed according to the property of SPGD AO system, the operate principle of each module was introduced specifically; all the voltages applied to the DM actors are computed parallel in order to increase the iteration time;the simulation results show that the iteration time of the software control system mentioned above can reach 362.88 us technically.A 21-element SPGD AO system for correcting wave-front distortion wasestablished, which used a PMT as a performance metric receiver. The experiment of correcting different type of wave front distortion, static and dynamic caused on purpose respectively is conducted on the system. Based on the logic analyzer, the iteration time of the SPGD algorithm practically is 399 us. The experimental results demonstrated that the SPGD AO system based on FPGA, can significantly improve the laser beam quality, correct the static wave-front distortion. The iteration frequency of the SPGD control system is 2500 Hz, the time of system stability is32 ms. As for artificial dynamic distortion, the system stability time can be 100 ms.There are many solutions improving the iteration speed, such as optimizing the optical path, using higher clock frequency, and outputting the control voltage multiply by using multiple D/A chips.In this paper, an adaptive optics system based on Hartmann control sub-system and SPGD AO sub-system cascade is established. The experiment of correcting wave front distortion is conducted on the system under circumstances of both sub-systems on, the Hartmann control sub-system off, the SPGD AO sub-system off,and both off. It turns out that, the cascade system is a remedy of application boundedness of each AO system, the success of the experiment inside the lab prepares the ground for astronomical observation.