Reliability-based Design Optimization Of Spacecraft Structure Based On Improved Surrogate Model

Stiffened shells and lattice grids are common structures in aerospace engineering,which are widely used in fuel tanks,fairings and interstages of launch vehicles.The task of lightweight design is particularly important for aircraft design.Furthermore,for spacecraft structure,there are a lot of uncertain factors,which come from the load environment,the material property,the manufacturing tolerance and the simplification procedure of calculation model,etc.Various uncertain factors can be taken into consideration with requirements of economy and security by using reliability-based design optimization.In this paper,a highly efficient and stable framework of global reliability-based design optimization is established for stiffened shell and lattice structures,based on an improved-precision surrogate model.The reduction of structure weight is taken as the goal.Optimization of the geometrical sizes and number of stiffeners of structure is completed with random factors such as these in geometry,material properties and load distribution,in which,the conditions of failure mode of buckling under axial force,bending moment,shear force and external pressure are mainly considered.There are four steps in the main content of this optimization framework.Firstly,establish an improved-precision surrogate model following the steps which are determining model variables by a sensitivity analysis,selecting the type of surrogate model,expanding the design space of training set,globally searching to optimize the parameters of the model and relaxing the precision requirement of local design domain.Secondly,obtain the random distribution of the complex load information and then transform it into normal distribution through equivalent normalize transformations.Next,obtain the reliability information using Adaptive Chaos Control Method in the inner layer,and transfer it to the global optimization algorithm in the outer layer.Finally,as the optimal design of the reliability optimization is required,re-establish a more accurate local surrogate model near the optimum point,and verify the failure probability using Monte Carlo Method.Then,the optimization framework is used to achieve the reliability optimization of two typical configurations of spacecraft,the metal orthogonally stiffened cylindrical shell structure and the composite Kagome lattice grid cylindrical and conical shell structure.It has been verified that the effectiveness and efficiency of this optimization framework are good enough to solve the reliability-based optimization problem for these engineering structures.And the universal law of improving the efficiency of bearing capacity and safety reliability of these structures is explored by changing the geometrical sizes.In addition,two ways are also put forward in considering the uncertainty of load in reliability-based design optimization: one is to allow the load value as a random variable subject to the actual load distribution,the other is to replace the allowable load value using a linear sum of its mean value and standard deviation.In the optimization framework presented in this paper,both of these two methods can propose a good lightweight design in the premise of meeting the structure safety reliability requirements,with effects better than the result from deterministic optimization based on 1.5 safety factor method.