Linear and Non-Linear Preprocessing of Wavefront Sensor Slope Measurements forImproved Adaptive Optics Performance

摘要

New methods for preprocessing wavefront sensor (WFS) slope measurements arepresented. Methods are developed to improve the accuracy of WFS slope measurements, as well as estimating key atmospheric and system parameters from the slope signals. Both statistical and artificial neural network solutions are investigated. Also, new atmospheric models for generating slope and phase data with the proper spatial and temporal statistics are developed. The experiments in improving the accuracy of WFS slope measurements include reducing the WFS slope measurement error and compensating for adaptive optics system time delay through temporal slope prediction. The experiments in key parameter estimation include estimating the Fried coherence length, r0, the wind speed profile, the strengths of the atmospheric turbulence layers, and the WFS mean square slope estimation error. Results of the experiments are used to make generalized conclusions in several key areas: first, the types of useful information that can be extracted from the WFS slope measurements; second, a comparison of linear or non linear methods; and third, the possibility of methods that can be developed which operate over useful ranges of seeing conditions. Overall, we find that the WFS slope measurements do contain useful information which can be extracted through various techniques. Simple transformations (either by neural network or statistical solution) on slope measurements can yield significant improvements is system accuracy without major changes to the adaptive optics system.