考虑随机变量的分布类型、功能函数的非线性程度、随机变量的变异性、随机变量之间的相关性等因素的影响,系统地对比分析了基于Orthogonal变换和Nataf变换的一次可靠度方法(FORM)的计算精度。研究表明:当随机变量的变异系数较小时,FORM的计算误差主要来源于Orthogonal变换或Nataf变换引起的非线性,而功能函数的非线性影响相对较小;同时,基于两种变换方法的FORM计算精度随着随机变量偏态系数绝对值的增大而降低;对于正态和对数正态分布类型,两种变换的计算精度相当,此时推荐采用Orthogonal变换;而对于极值Ⅰ型分布和移位瑞利分布类型,当随机变量的变异性和随机变量之间的相关性较小时,两种变换的计算精度相当;反之,当随机变量之间高度相关时,Orthogonal变换的误差较大,而Nataf变换的计算精度较好,此时推荐采用Nataf变换。进一步的研究表明,基于Orthogonal变换和Nataf变换的FORM同样适用于具有隐式功能函数的刚架结构,且两种方法的计算精度相当。 Considering the distribution types of random variables, the degree of non-linearity of function, the variability of random variables and the correlation between random variables, a method of reliability based on Orthogonal transform and Nataf transform ) Calculation accuracy. The results show that when the coefficient of variation of random variables is small, the calculation error of FORM mainly comes from the nonlinearity caused by Orthogonal transform or Nataf transform, but the nonlinear effect of function function is relatively small. Meanwhile, the FORM calculation based on two kinds of transformation methods The accuracy decreases with the increase of the absolute value of the skew coefficient of the random variable. For the normal and logarithmic normal distribution types, the calculation accuracy of the two kinds of transformations is equivalent, Orthogonal transform is recommended at this time; The displacement Rayleigh distribution types, when the correlation between the variability of random variables and the random variables is small, the accuracy of the two transformations is comparable; on the contrary, when the correlation between the random variables is highly correlated, the error of the Orthogonal transform is larger, The Nataf transform calculation accuracy is better, then recommended Nataf transform. Further research shows that FORM based on Orthogonal transform and Nataf transform is also suitable for rigid frame structures with implicit function, and the computational accuracy of the two methods is equivalent.