首先引入等概率边缘变换的基本原理,证明了常用的Rackwitz-Fiessler变换是等概率边缘变换的一次近似形式,将当量正态化原理和线性变换相结合,提出了扩展的Rackwitz-Fiessler变换,并指出其存在的缺点。然后针对Nataf变换的非线性特征,提出了线性化Nataf变换,并将该变换与改进的HLRF算法相结合,给出了基于线性化Nataf变换和iHLRF算法的一次可靠度方法。将Nataf变换、线性化Nataf变换和扩展的Rackwitz-Fiessler变换通过算例进行了对比分析,结果表明:采用线性化Nataf变换的结构可靠度分析结果收敛于采用Nataf变换的计算结果,而采用扩展的Rackwitz-Fiessler变换的计算结果则有较大的误差。
The principle of iso-probability marginal transformation was firstly introduced, and then, it was proven that the usual Rackwitz-Fiessler transformation is the first order approximation of the iso-probability marginal transformation. The extended Rackwitz-Fiessler transformation was developed through the combination of the equivalent normalization principle and linearized transformation, and its defects were also pointed out. Noting that Nataf transformation is nonlinear for non-normal variables in nature, so a linearized Nataf transformation has been put forward. Furthermore, a new FORM method based on the proposed linearized Nataf transformation and iHLRF algorithm is proposed. Finally, the computing results of a numerical example are analyzed and compared between Nataf transformation's and linearized Nataf transformation's and extended Rackwitz-Fiessler transfor- mation's, respectively. The example shows that the results of structural reliability analysis based on linearized Nataf transformation convergence the ones resusting from Nataf transformation, while there are errors in the results of extended Rackwitz-Fiessler transformation by comparing with the exact ones.