| The casing of a solid rocket motor is an important component of the motor and directly affects its performance.However,due to the complex structure of the casing,as well as uncertainties in size,material,and environmental factors,traditional stress-strength interference theory is difficult to directly apply to solving the response surface equation of the casing’s limit state.This paper investigates the use of finite element analysis methods to explore the impact of different response surface construction methods on the reliability analysis of solid rocket motor casing structures.By comparing the use of different response surface construction methods to calculate reliability indicators such as failure probability,the impact of response surface construction methods on reliability indicators is determined.A faster and more accurate limit state equation simulation method is constructed using the stochastic response surface method,which replaces traditional finite element analysis methods.This method can calculate reliability indicators of solid rocket motor casing structures more quickly and with higher accuracy.By comparing the reliability indicators calculated using the finite element analysis method and the limit state equation simulation method,the accuracy and reliability of the limit state equation simulation method are verified,providing reference for engineering practice.Based on mechanics theory and finite element analysis results,it can be seen that the maximum equivalent stress experienced by the solid rocket motor is located at the opening of the front end.Based on this,a solid rocket motor reliability calculation method based on PCE stochastic response surface method is proposed: first,the metamodels of response surface,stochastic response surface,and other proxy models are established based on response results obtained through LHS sampling.Through comparative analysis of various proxy models,it is found that the traditional quadratic response surface has lower accuracy,while the stochastic response surface has higher accuracy.Other proxy models such as neural networks and regression tree methods have their own advantages and disadvantages and need to be selected based on different engineering situations.Based on PCE regression method,the least squares method,moving least squares method,and an improved moving least squares method are compared.The moving least squares method considers the weight of sample points near the design point,and other sample points also participate in the construction of the regression equation,considering both global and local accuracy.The improved moving least squares method further considers the weight of sample points to the design point and response surface,further improving the calculation accuracy.Then,the reliability index of the PCE response surface is obtained using the maximum entropy quadratic fourth-order moment method,and the probability density function of the limit state equation is constructed.The maximum entropy quadratic fourth-order moment method is affected by the first four moments of the response variable.By comparing the first-order second-moment method,the optimal square approximation quadratic fourth-order moment method,and the maximum entropy quadratic fourth-order moment method,the reliability index of the response surface is obtained.Finally,when considering the impact of time variables on a single random variable,it can be seen that as working time increases,the structural reliability also changes.Therefore,time variables are also an important factor in the reliability calculation process. |
