中文摘要：

在回归模型研究与应用中，非正态分布假设、非线性回归模型与组合模型建模是重要的研究问题。首先针对非线性建模，引入压缩函数，提出了一种非线性回归建模方法；其次针对单个模型的局限性，引入线性回归与时间序列模型组合，提出了一种组合加法模型的建模方法；最后针对非正态分布假设建模，利用随机变量和分布，提出了一种基于非正态分布的和模型，为非线性回归、组合模型与非正态分布假设建模提供了新的方法。

英文摘要：

In the study and application of linear regression model, modeling methods based on abnormal distribution, non-linear regression and combined models are three important problems. In order to solve these difficult problems, a new method of building non-linear models is firstly proposed by applying functions. A method of building additive models is then put forward by combining linear regression and time series models. Finally, a modeling method from abnormal distribution is advanced by applying summation distribution in the paper.