Structural Reliability Analysis Based On Multiple Surrogate Model

Commonly used reliability analysis methods in practical engineering problems,such as the first-order reliability method and the classical response surface method,they both have shortcomings on calculation accuracy.Although Kriging model is introduced into reliability analysis and solves some problems above,there are still some high dimensional and high nonlinear problems which need greater computational cost.In view of these problems,some related research has done in this paper,as follows:(1)This paper combines the advantages of classical response surface method with those of Kriging method.At the same time,a new strategy of selecting sample points is designed and applied,and then a multiple surrogate model is proposed.The concept of reasonable sub-region is introduced into space of basic random variables,and the space is divided into two independent regions.Different surrogate model is used to calculate reliability in these two different sub-regions so that obtain more precision calculation results.(2)In the space of basic random variables,the regions which contribute a lot to the failure probability of structures are included in the reasonable sub-region.And the reasonable sub-region is defined as a hypersphere with the most possible failure point as the center of the sphere.According to the corresponding relationship between reliability index and failure probability of structure in standard normal space,the radius calculation formula of reasonable sub-region is deduced.(3)Through analyzing two-dimensional limit state problems,the operation mechanism of the proposed method can be observed visually.And visualized image of the calculation results is drawn.It can be seen from the image that the multiple surrogate model proposed in this paper has higher precision of calculation.Through analyzing high-dimensional and non-linear problems,and the calculation results are compared with those obtained by other methods,which shows that the method in this paper has better calculation precision and lower calculational cost.