证明了离散齐次指数函数经一次累加生成后为离散非齐次指数。进而在GM(1,1)以均值生成作为灰导数背影值的基础上 ,进一步提出了一种逐步优化灰导数背景值的方法 ,提高了建模精度 ,特别对于绝对灰度为0 (或很小 )的具有齐次灰指数律的数据 ,应用该方法可以得到十分理想的预测模型。 It is proved that the discrete homogeneous exponential function is a discrete inhomogeneous index after a cumulative summation. Furthermore, based on the mean value of GM (1,1) generated as the gray derivative backscattering value, a method of gradually optimizing the gray derivative background value is proposed, which improves the modeling accuracy, especially for the absolute gray scale of 0 Small) with homogeneous gray exponential law, the application of this method can be very good prediction model.