Multidisciplinary Design Optimization Using Dynamic Metamodel For Earth Observation Satellites

Satellite conceptual design is a complicated system engineering involving a number of disciplines.Conventional design mode fails to consider the intercoupling among different disciplines which largely restricts the design level.With the increase of space launch missions and improving requirements for high performance spacecrafts,more and more drawbacks of the conventional design mode has been exposed.Multidisciplinary design optimization(MDO)technology to improve the design quality and efficiency has become the trends of satellite overall design.Earth observation satellite can obtain detailed data and information of the earth,which plays an important role in the national economy,social development and national security.Then earth observation satellite technology is essential for the development of all countries in the world.Therefore,aiming at the earth observation satellite MDO problems,the paper engages in the researches from two main aspects consisting of satellite MDO modeling method and the dynamic MDO strategy based on metamodel technology.The main contents are summarized as follows:(1)Overview of the development of the satellite conceptual design,the-state-of-the-art of the key of thchnologies of multidisciplinary design optimization and research progress of multidisciplinary design optimization for satellites.(2)Taking China Brazil resources satellite as an example,the design variables,constraints and modeling analysis conditions of orbit,payload,power,attitude control,orbit control and structure disciplines are considered,and then the data flow of earth observation satellite conceptual design is studied.At last,the MDO model of earth observation satellite is established to maximize the satellite observation performance,which provides the foundation for following work.(3)The dynamic metamodel technology has been studied based on the brief introduction of the commonly uesd experimental design method,metamodel method and constraint processing method.Aiming at the limitation of efficient global optimization method(EGO)to solve constrained optimization problems,an adaptive penalty function is used to adjust penalty coefficients to deal with high precision constraints(When feasible solution satisfying constraints is obtained,the penalty coefficient is reduced so that the next optimization iteration is more likely to improve the optimality,on the contrary,the penalty coefficient is increased so that the next optimization iteration is more likely to improve the feasibility),and the efficient global optimization strategy based on adaptive function(C-EGO)is proposed.The comparison of the performance of C-EGO with the constraint important mode pursuing sampling method(CiMPS)and the genetic algorithm(GA)shows that C-EGO has obvious advantages in convergence and optimization efficiency through standard test cases.(4)A discrete variable approximation optimization strategy based on dynamic radial basis function metamodel(DTR-DRBF)is proposed for the optimization of discrete variables in satellite design optimization.By introducing the parameter mapping technique,the discrete variables are mapped into integer space,then the metamodel is constructed by integers,and the metamodel is optimized by integer genetic algorithm until convergence.In addition,the sequential Maximin latin hypercube design method is used to enhance the spatial uniformity and the projection uniformity of the sampling results,thus improving the efficiency of the algorithm.The comprehensive performance of DTR-DRBF is tested by standard engineering example,which proves the validity and engineering practicability of the method.(5)A multidisciplinary design optimization model for earth observation satellites was established.The C-EGO dynamic optimization strategy proposed in this paper was used to solve the multidisciplinary optimization problem.The conceptual design scheme was obtained and the effectiveness of the optimization strategy was proved.Then a composite satellite structure optimization model was established based on the above design scheme.The optimum design was achieved by using DTR-DRBF,and the optimal solution satisfying the constraints was obtained.The quality of the satellite center bearing cylinder was reduced by 26.68%.