Optimal Control Technology Of Adaptive Optics

Adaptive optics(AO) systems can compensate random distorted wave-fronts caused by atmospheric turbulence and internal disturbance of optical systems, its control process mainly comprises deformable mirror control and tip/tilt mirror control. Optimal control technologies of AO systems are effective ways to develop ultimate correction abilities. However, conventional AO systems with classical proportional and integral control methods often struggle to correct broadband atmospheric turbulence and high-frequency vibrations due to system latency and corrector’s characteristics. Aiming at above problems, the main contents of this dissertation are summarized as follows: optimizing the modes of beacons and wave-front detecting to reduce time delay fundamentally; synthesizing a vibration controller to optimize the control loop of the tip/tilt mirror; proposing a double overlapped AO system architecture and optimizing its operate patterns; proposing a wind predictive control based on wind estimation to increase control bandwidth.AO systems need beacons to detect wave-front information, different modes of beacons and wave-front detecting have a great influence on the performance criterion such as the system bandwidth and stability. Researching the performance of the system is full of critical meaning to properly design and evaluate a system. Based on the system work flow, the light wave characteristics of beacons and the methods of exposure and read-out of wave-front sensors, the transfer function models of AO system under four operate modes including continuous light frame transfer, continuous light line transfer, laser pulse frame transfer and laser pulse line transfer, are built using mechanism analysis methods. The calculation equations of system bandwidth and phase margin are derived. The effects of sample frequency and controller parameter on system performance are studied. Results of analysis show different modes of beacons and wave-front detecting cause different system time delays, which makes system bandwidth and stability differ.Optical axis jitters, which arise from different factors such as wind shaking and structural oscillations of optical platforms, have a deleterious impact on the performance of AO systems. When conventional integrators are utilized to reject such high frequency and narrow band disturbance, the benefits are quite small; even worse, the system may fail to operate normally due to amplification. On the basis of observed data, its frequency characteristics are analyzed. With the help of the Smith predictor, one effective and stable technique to design a controller is proposed on account of the peak frequency and bandwidth of the jitter. The relationship between controller parameters and filtering features are discussed, and the robustness of the controller against changing parameters of the control object is investigated. Results show the variance of one axis aberrations caused by optical axis jitters can be reduced by about 60% as a result of exploiting novel controllers, which compensates the deficiency of conventional controllers. Preliminary experiments are carried out to validate the algorithm. Results show expected filter responses can be acquired after simple system parameter identification.Optical systems like telescopes are constantly affected by atmospheric turbulence and internal aberration disturbance. Conventional AO systems are proved to be an effective way to eliminate external aberrations caused by atmospheric turbulence. However, when internal aberrations with high amplitudes emerge, the performance of conventional AO system is unsatisfactory. Hence, two suits of AO system are employed to operate jointly. This paper presents a double overlap AO system structure, which exploits an independent inner channel AO system to correct internal aberration disturbance, and an outer channel AO system to mainly correct external atmospheric turbulence. Furthermore, the correction abilities of this structure and conventional AO system are analyzed and compared. In addition, their correction performances under the situation where two aforementioned type aberrations simultaneously exist are numerically simulated. Results show that the ability to reject internal aberration disturbance of DOAOS is near two times greater than conventional AO systems. Under the operation condition where the SNR of the inner channel is high enough and the SNR of the outer channel is extremely low, the residual variance of DOAOS is 10%~50% smaller than conventional AOS when internal aberration disturbance is relatively strong.The applied correction lags behind evolution of distorted wave-fronts as a result of system latency caused by read-out and control computation, which sets a limit to the system performance because of inadequate correction bandwidth. Predictive control methods are critical means to increase correction bandwidth other than increasing the sampling frequency. According to frozen flow turbulence hypothesis, horizontal wind dominates the main trend of atmospheric turbulence evolution. So it is useful to reduce the effect of time delay by predicting atmospheric turbulence with wind estimation. In this dissertation, a wind predictive control based on wind estimation is researched, which mainly comprise real-time wind estimation through optical flow computation and wave-front prediction. The wind estimation method is tested by using slope measurements and reconstructed voltages from 127 unit AO system, respectively. The latter is experimentally validated. Results show that relative error for the estimated wind magnitude is almost 10%, and the error for wind direction is within 10 degree. The wind estimation method is applied to wave-front control process in open loop and closed loop architecture for 127 unit AO system. Simulation results show that wind predictive control can decreases the residual variance acquired by a delicately designed conventional controller by about 10%.In this dissertation, the analysis of beacon and wave-front detecting modes as well as double overlapped AO system provides foundations for system design and performance evaluation. The vibration controller is of high realistic value. Results of preliminary numerical simulations demonstrate that wind predictive control based on wind estimation have better rejection ability than conventional proportional integral control under the condition of large wind speed. In the near future, experiments will be taken to evaluate wind predictive control algorithm.