| As the optical design systems have developed rapidly before,the aspheric optical elements have been used widely in Aerospace,laser guidance,lithography and many other aspects.Based on the sustained progress of manufacturing and computer numerical control processing technology,the accuracy and efficiency of aspheric were improved critically.But there are still defects in the aspheric test,and constrain its further application.The null-test Offner compensation in interferometry has the advantages of high resolution,high stability,high sensitivity,etc.It can be used as the benchmark of selfinspection or comparison test,and is the main detection method of surface shape error in aspheric surface precision polishing stage.The Offner compensator is composed of two or more lenses.Most of the existing design methods are based on the third-order aberration theory,aiming at a specific form to solve the null compensation system.Because the compensator corresponds to the tested aspheric surface,the universality of the compensation system is insufficient.It is difficult to find the complex relationship between the parameters of the system and the inspection ability,and it is difficult to investigate the null test ability of the compensation system.Moreover,there is little research on the relationship between the effective aperture,incident angle aperture,exit angle aperture,lens power and the ability of compensating optical path.There is no clear and intuitive understanding of the influence of structural parameters on the inspection effect in the optical path.The design is blind and often depends on the designer’s goal and experience.The evaluation of the results is usually not unique,and there is no data for reference on the optimal design results that can be achieved by a compensator with specific optical parameters and structural forms.This paper is dedicated to figure out the generalization problem of paraboloid Offner compensation inspection and evaluation,and on this basis,it provides a new idea for optical design evaluation.Specifically,the compensation of residual wave aberration is taken as the evaluation index,and the vertical axis magnification and the relative aperture of the tested aspheric,which are positively related to the compensation ability of the compensator,are selected as the parameter space variables.A method of evaluating the design results of paraboloid null-test Offner compensator based on artificial neural network algorithm is proposed.The powerful nonlinear information processing function of neural network is used to train BP neural network,and a large number of excellent design results are used as the data basis for learning and reasonable extrapolation,so that the complex relationship between structural parameters and design compensation residuals is found and reflected in the form of data and charts.The user interaction interface is established.According to the user’s self-defined input,a prediction model can be given,which helps designers to judge the optimization direction and choose a reasonable model when optimizing the system.By establishing the database of the ultimate optimization results of the tested aspheric and the compensators corresponding to several structural forms,the evaluation reference diagram of design capability is given,which can provide prior design standards for compensation design indexes.The main functions are as follows:Firstly,a method of using neural network to store and manage evaluation data is realized.In the design of Offner compensator,similar surface error results can be obtained by changing different structural parameters under certain conditions,and the optimal structure needs to be judged on the basis of experience,so there are many uncertain interference factors,which takes time and effort.In this study,the perfect design in the standard design parameter space of Offner compensator is realized,and the neural network tool is used for management,which helps designers to set aside targeted adjustment space when the existing design model has not reached the target requirements,and to evaluate the excellent level of the model after the optimization is completed,so as to improve the efficiency and reduce the trial and error cost.Secondly,the capability evaluation work under several design degrees of freedom is compared with the standard evaluation data,and the design map is obtained,and the function of updating the design data is realized in the software.In the design of off compensator,there is a nonlinear mapping relationship between structural parameters and evaluation indexes,and the corresponding influence weights are not expressed qualitatively.Based on certain data analysis,this study points out the contradiction between structural parameters and evaluation quantities,gives a design evaluation reference map,and updates and perfects it when the design result input by the user is better than the model level in the established database,so that the evaluation reference map can be iterated continuously. |
