本文提出了一类边际分布为加法危险率模型的copula模型来研究多维生存时间变量之间的相关关系。主要对边际分布为常系数加法危险率模型的Clayton copula模型展开研究，采用两阶段估计法得到该模型参数的估计。具体地，第一阶段通过Lin等提出的方法得到边际生存函数的估计；第二阶段将估计出来的边际生存函数带入copula似然函数中得到伪似然函数，然后使用Newton-Raphson算法极大化伪似然函数得到copula常联结参数的估计。当联结参数为某个协变量的函数时，采用基于B-样条技术的两阶段估计法计算copula变联结参数的估计。此外，还证明了变联结参数估计量的大样本性质，并结合Bootstrap方法进行随机数值模拟实验，验证了该估计方法在有限样本下的表现。

In this paper, a copula model with marginal distribution as additive hazard model is proposed to study the correlation between multidimensional survival time variables. The Clayton copula model with marginal distribution as constant coefficient additive hazard model is mainly studied, and the two-stage estimation method is used to obtain the estimation of the model parameters. Specifically, the first stage obtains the estimation of the marginal survival function through the method proposed by Lin et al.; in the second stage, the estimated marginal survival function is brought into the copula likelihood function to obtain the pseudo-likelihood function, and then the Newton-Raphson algorithm is used to maximize the pseudo-likelihood function to obtain the estimation of the copula constant linkage parameter. When the linkage parameter is a function of a certain covariate, the two-stage estimation method based on B-spline technique is used to calculate the estimation of the copula variable-linked parameter. In addition, the large sample nature of the variable-linked parameter estimator is also proved, and the random numerical simulation experiment combined with the Bootstrap method is carried out to verify the performance of the estimation method under a limited sample.