Structural Reliability Analysis And Parameter Sensitivity Analysis Of Rocket Tank Based On Surrogate Model

With the progress of China’s manned spaceflight and space exploration,the frequency of launch vehicles in China is increasing,and the high-reliability of hull structure has been one of the core concerns of launch vehicle system engineering.Based on the reliability of the launch vehicle tank structure,this paper studies the reliability analysis method of the Kriging surrogate model based on active learning and the sensitivity analysis method of the RBF surrogate model based on active learning under the premise of ensuring accuracy.It is applied to the engineering of launch vehicle tank structure.The main research contents of the thesis are as follows:(1)The Kriging surrogate model method is a relatively mature method applied in the field of structural reliability calculation.Based on this theory,combined with some active learning methods,the Kriging surrogate model has been continuously improved.The research combines the method of active learning function Kriging surrogate model to solve the problem of structural reliability,and uses classic examples to verify the universality of this method in structural reliability problems of different dimensions,linear nonlinearity,and implicit state surfaces.(2)The related content of radial basis function is introduced,and the parameter sensitivity of structural response is calculated based on the radial basis function(RBF)surrogate model combined with the active learning method.In this paper,the RBF active learning surrogate model is constructed,and the method for solving the parameter global sensitivity analysis index based on variance is given.Considering the uncertainty of random variables,the feasibility of the method is verified based on the truss structure calculation example.(3)Taking a certain type of launch vehicle propellant tank as the research object,the structural reliability and sensitivity are studied,and the reliability and parameter sensitivity analysis methods proposed in this paper are applied to engineering.Based on ABAQUS,a refined parameterized model of the rocket tank structure is established,and the reliability of the tank structure is solved based on the active learning Kriging surrogate model method.Taking the load-bearing capacity of the tank structure as a response,based on the method of active learning of the RBF surrogate model,combined MATLAB and ABAQUS to solve the global sensitivity under parameter uncertainties.Quantitatively evaluate the reliability and sensitivity of the tank structure,provide data support for engineering applications and give reasonable suggestions.