Modeling and Control of a Nonlinear Dynamic Adaptive Optics System

Closed-loop adaptive optics (AO) systems are widely used in many scientific and medical applications, such as astronomy, laser systems and microscopy, to improve the resolution of the image by actively sensing and correcting the optical aberration in the system. The performance of the closed-loop AO system with conventional linear-static-model based controller degrades if the AO system suffers from nonlinearity or the dynamics is significant. In this paper, we propose and validate a strategy to model and control a nonlinear dynamic AO system. The performance of the closed-loop AO system is improved in three successive steps: hysteresis compensation, closed-loop subspace identification and model-based controller design. Experimental results show that the convergence speed of the closed-loop AO system has been improved significantly. The variance of the residual error has been reduced by 30% with respect to the conventional AO control approach.