基于AK-SDMCS的含缺陷核管道可靠性分析

Reliability analysis of flawed nuclear piping based on AK-SDMCS

为克服含缺陷核管道可靠性分析面临的计算效率低、精度差等问题,本文基于参数空间球分解思想提出了一种改进的主动学习Kriging代理模型方法。首先,给出含环向内表面缺陷核管道的主要失效模式和随机参数,并建立极限状态方程;其次,基于球空间分解思想改进主动学习Kriging模型方法,克服现有方法在求解小失效概率时主动学习本身计算量过大的缺点;最后,将所提方法应用于含缺陷核管道可靠性分析,并通过3个算例进行验证。结果表明,所提方法具有较高的计算精度和效率,可为含缺陷核管道可靠性分析提供有效手段。

In order to overcome the disadvantages such as inefficiency and inaccuracy encountered in reliability analysis of flawed nuclear piping,this paper proposes an improved active learning-Kriging metamodel method based on the idea of spherical decomposition of parameter space.First of all,the primary failure modes and stochastic parameters of nuclear piping containing circumferential internal surface defects are given,and the limit state equation is built.Then,the active learning Kriging metamodel method is improved based on spherical decomposition idea to tackle the shortcoming of high computational cost of active learning itself when computing small failure probabilities.Finally,the proposed method is applied to reliability analysis of flawed nuclear piping,which is demonstrated by three numerical examples.Results show that the proposed method exhibits high efficiency and accuracy,which provides an effective means for reliability analysis of flawed nuclear piping.