Research On Analysis Of Mirror Deformation And Structural Optimization For Protective Window Of Large Aperture Theodolite

Photoelectric theodolite is a observing and measuring device.Its main functions are to track targets and afford highly clear and intuitionistic real-time images.The protective window is installed before the main system of theodolite for avoiding the corrosion of salt fog,damp and mildew.But the deformation of protective window will result in the change of wave front aberration,influencing the imaging quality of optical system.So deep research is done in this paper.In order to analyze the mirror deformation of meter class theodolite protective window under exterior loads,the model of meter class theodolite protective window is built on the base of frictional contact finite element theory.And the differences of linear rigid joint and nonlinear frictional contact simulation methods are compared.Homogeneous transformation is used to eliminate rigid displacement,from which distortion of meter class theodolite protective window are gotten.Zernike polynomial is applied to the interface tool of integrated analysis,and Zernike coefficients are used in Zemax to get wave front RMS.When the angle of pitch of protective window is 0 degree,the RMS and PV of protective window surface are 38.095 nm and 205.027 nm considering frictional contact,and the values are 40.626 nm and 235.654 nm by interferometer testing.The results show that the deviation between simulated and experimental results is 6.23%,so the simulation considering contact boundary can reflect the mirror deformation of meter class protective window more precisely.The effect of pulsating wind on structure has uncertainty.In order to eliminate the effect of pulsating wind on the system imaging quality,the mirror deformation prediction of theodolite protective window under pulsating wind is needed.This paper provides a method,which could simulate the nonlinear relation of pulsating wind press and mirror wave front RMS,called NARX neural network.This method applies AR module to simulate the time sequence of pulsating wind speed,and mirror deformation data is gotten by using finite element analysis.These data are used for the training,validation and prediction of NARX neural network offline.The result of simulation shows that the built NARX neural network predictive MSE is 3.4348×10^-9,and it can satisfy the real-time requirement of active compensation of theodolite system mirror deformation.The thickness of structure and glass of protective window for meter class aperture theodolite is optimized by BP neural network.The optimized thickness result is 18.5mm and 83.5mm.The reliability of the design of protective window is guaranteed in the largest degree.