Research On Structure Optimization Of Large Aperture Monolithic Space-based Mirror And Its Mounting Technology

With the continuous development of space optical technology,the requirement of the resolution observing the earth became higher and higher.The important way to improve observing resolution was increasing the aperture of the optical sysytem.So the research of large aperture space-based mirror was in an unprecedented hot state.The large aperture space-based monolithic mirror for earth observation had some advantages,such as no segment asembly,high surface figure accuracy and relatively simple supporting system and so on.Therefore,it bacame the hot study spot far all engineers.According to the empirical formular between the mass and the diameter of mirror,the mass increased in proportion to the diameter to the third power.The increased mass of mirror may raise launch costs largely for space sensor,and complicate the supporting system.Also,the increased mass of mirror may reduce characteristic frequency of the mirror assembly and increase its dynamic response.This paper presented a prediction model based on artificial neural network for optimizing mirror structure parameters.In order to obtain the nonlinear mapping relation between target characteristic and design variables for large aperture mirror structure optimization,we built a prediction model based on Back Propagation(BP)neural network.First,defined the monolithic central thickness,rib thickness,face sheet thickness and outermost wall thickness as network input parameters,and the output parameters of neural network included weight of mirror,first order constraint mode and the surface shape Root Mean Square error under axial gravity.Training samples and test samples of neural network were obtained by taking advantage of orthogonal test design and finite element analysis software Patran/Nastran.Taking advantage of Matlab tool,we adjusted transfer function and the number of hidden layer neurons and training algorithm to build a prediction model of mirror target characteristic that meet error requirements.And the model can simulate the nonlinear mapping relation between target characteristic and design variables.Extrapolation performance of this prediction network was validated by test samples.The results indicated that,its relative error were under 3.83% to predict mirror weight and first order constraint mode with this neural network prediction model,and its relative error were under 7.09% to predict surface shape Root Mean Square error under axial gravity using this model.High precision extrapolation performance of network model ensured optimization of structure parameters.Finally,the structure parameters of mirror were optimized with this network model and attain a mirror structure with certain parameters.The most important consideration of supporting technology for large aperture mirror was its precise position.This paper presented the exact constraint theory that used in precision instrument design detailedly,and proposed a method to design the exact constraint line diagram for mirror with dual line rule.Then designed three schemes to mirror exact constraint using this method.Also,designed flexible support structure on the basis of degree of freedom in the constraint line diagram one by one.Using software of finite element analysis,analyzed statics,thermal adaptability,the first three modal frequencies and safety margin of the three schemes mirror assembly.Finally,this paper selected the third flexible support structure as positioning support by considering simulation data,the convenience of carrying out gravity compensation and its occupied space overrall.For mirror assembly with flexible support structure,thermal adaptability and its first order modal frequency were mutually contradictory target properties.We not only hoped that the mirror assembly had high first modal frequency enough but also hoped the mirror assembly had well thermal adaptability.This paper proposed an efficient and convenient method to multiparameter optimization for mirror flexible support structure.Built optimization process by calling pre-processing and post-preprocessing software of finite element analysis Patran,software of finite element analysis Nastran,surface shape error procesing software Matlab and so on with multi-disciplinary optimization software Isight.Automatically modified model parameters on the basis of the results using optimization algorithm,and automatically carried out iterative calculation until the objective was optimal.First,obtained the relation schema between the target properties,i.e.first order modal frequency and the surface shape Root Mean Square error under 3 degree centigrade rise,and each input parameter.Then,with the constraint condition that the first order modal frequence was greater than 45 Hz,obtained the combination of input parameters that makes the mirror had optimal thermal adaptability.The comparison of initial design and optimization result showed that the surface shape Root Mean Square error under 3 degree centigrade rise of mirror droped 31.6 percent after opitimization,and the first order modal frequence remained unchanged.For large aperture space-based mirror,the biggest difference of mechanical environment between on-ground and on-orbit was gravity.On ground,large aperture mirror needed gravity compensation to guarantee the same or similar mechanical environment between on-ground and on-orbit.This paper used the method that installing force actuator under the mirror to compensate gravity.Designed a mechanical actuator scheme that could output constant force to mirror gravity compensation.Through elabrorate finite element analysis,it was proved that 51 compensation support points were needed to make sure the surface shape Root Mean Square error is below 6nm.Finally,the force magnitude of each compensation support point was caculated by the mothod of reverse constraint calculation,that was constrained all compensation support point in the FEM model and found these constraint force when the surface shape Root Mean Square error meeted requirment.These constraint forces were compemsation force we needed.And then verified these force by means of applying these force to the mirror FEM model,the result showed that the surface shape Root Mean Square error could reach 5.54 nm under 51 compensation support points.The rationality of the method was verified.