基于环形抽样代理模型的结构可靠性分析方法

该文提出一种基于高效代理模型—Kriging模型的可靠性分析方法，通过欧式距离构造抽样区域，用于确定代理模型更新过程中样本点的选取范围，保证代理模型在更新过程中只对失效概率贡献大的区域选取样本点。算例分析表明该方法具有较高的精度和效率，不仅适用于性能函数为隐式的结构可靠性分析，而且通过改变环形抽样区域的带宽，同样适用于结构可靠性分析中的小概率失效问题，为结构可靠性分析提供了新途径。

Structural Reliability Analysis Using Circular Sampling Surrogate Models

The performance functions of components or systems are often implicit in engineering. A new reliability analysis method based on efficient surrogate model, Kriging model, is proposed. The sampling area is determined by the Euclidean distance and used to determine the area of sampling in the update process of the surrogate model. The proposed method can sample just in the area which have more contribution to the calculation of the failure probability in the update process of the surrogate model. The numerical example shows that the proposed method is accurate and efficient, and can be used in the structural reliability analysis with implicit performance functions. Also, the proposed method can be applied to structural reliability problem with small failure probability by changing the width of the circular sampling area. The proposed method provides a new way to the structural reliability analysis.