中文摘要：

GM（1,1）模型的白化解为齐次指数形式,而一般数据呈非齐次指数形式,存在形式上的差异.本文运用非齐次级比与非齐次指数函数的对应关系,对原始序列中相邻数据做差处理,得到新的序列,将非齐次指数序列转换为齐次指数序列,再建立GM（1,1）模型.实例表明,运用初值优化和非齐次化能提高GM（1,1）模型的精度.

英文摘要：

The form of whitening solution of while that of general data is non-homogeneous GM（1, 1） model is homogeneous exponential exponential, there are differences in two forms In this paper, we firstly apply the corresponding relationship between the non-homogeneous class ratio and non-homogeneous exponential function, and do difference processing to the original sequence adjacent data to get a new sequence, which change the non-homogeneous exponential sequence to a homogeneous one, then establish a GM（1, 1） model.The application example shows that the fit precision of the initial value modify model and non-homogeneous grey model can be enhanced.