Study On Off-axis Telescope Misalignment Detection Technology

In the optical system,compared with the refractive system,the reflective system has a series of advantages,such as more compact optical path,larger field of view,wider spectrum and easier lightweight design.Compared with the axisymmetric reflective system,the off-axis reflective optical system does not have the advantages of central occlusion,high light energy utilization,no diffraction limit at medium and low frequencies,so it is widely used in the field of space optical communication,astronomical detection and space remote sensing.However,due to the change of external environment(such as temperature change,airflow disturbance and gravity effect)and self-dithering caused by mechanical structure of system,the spatial position of optical lenses will change,which will affect the overall imaging quality of optical system.However,with the development of processing technology,lens shape and mechanical structure of system are no longer the primary decisive factors limiting imaging quality of optical system,but misalignment detection has become one of the main factors affecting imaging quality of off-axis reflective optical system.In this paper,focusing on misalignment detection of off-axis reflective telescope system based on traditional sensitivity matrix method,second-order sensitivity matrix method,fully connected neural network method and convolution neural network method,the corresponding research on this specific engineering problem is carried out.(1)The application background,the process of initial alignment and the development of fine tuning technology of off-axis telescope are described.The distribution of aberration in off-axis system is deduced from aberration distribution of coaxial system by using vector wave aberration theory.Then,the aberrations distribution of off-axis system with misaligned error is deduced.The principle of current mainstream method of error detection is introduced,and their advantages and disadvantages are analyzed.(2)The accuracy of sensitivity matrix method will reduce when the misalignment error is large,the reason of sensitivity matrix ill conditioned is discussed by using the off-axis telescope model.It is found that the partial misalignments are nonlinear with Zernike polynomial coefficients.To solve this problem,a second-order sensitivity matrix method is proposed.In order to explore the feasibility of this method,the sensitivity matrix method and the second-order sensitivity matrix method are used to carry out simulation experiments.The results show that the second-order sensitivity matrix method has higher accuracy and wider application scope under the premise of larger misalignments.(3)In view of shortcomings of the second-order sensitivity matrix method in dealing with the coupling of a variety of misalignments,and because of strong nonlinear fitting ability of neural network,a fully connected neural network misalignment detection method is proposed.The simulation results show that the method can predict corresponding misalignment of system with different field angles,and has satisfactory accuracy.It is speculated that this method has a great application prospect in the field of off-axis telescope system alignment.(4)The common defects of fully connected neural network,second-order sensitivity matrix and sensitivity matrix method are summarized,that is,additional wavefront detectors are needed to obtain Zernike polynomial coefficients.To solve this problem,a convolution neural network misalignment detection method is proposed,which directly uses system speckle pattern to solve misalignments.With the help of Stewart six degree of freedom platform,a secondary mirror displacement platform is designed,and an experimental platform is built based on the secondary mirror displacement platform.Then the neural network is trained with the collected pictures,and the network can converge.The test results show that this method is feasible.