基于随机振荡序列离散灰色模型的优化

Optimization of Discrete Grey Model Based on Stochastic Oscillation Sequences

文章探讨了随机振荡序列灰色建模的问题,分析了两种具有代表性的基于振荡序列灰色建模方法,指出了其函数变换的实质以及应用范围。现有实证给出的建模方法其模型输出序列呈现出下凸的(除第一点外)和"Z字型"特点。结合这两种建模方法的优势,提出了另外两种数据变换方法,即加权均值平滑变换和级比调节平滑变换,实例说明了新的两种数据变换方法能有效地提高建模精度。

This paper discussed the grey modeli ng problem of stochastic oscillation sequences and analyzed two kinds of representative grey modeling methods based on the oscillation sequences. Besides,the essence and the application scope of two function transformations were pointed out. The empirical results shows that the existing modeling methods are presented in which the model output sequence exhibits lower convexity（except for the first point）and the ＂Z＂feature. By combining with the advantages of these two modeling methods,two other data transformation methods are proposed ：the weighted average smooth transformation and the class ratio-adjusted smooth transformation. In conclusion,the example also shows that the two new data transform methods can effectively improve the modeling accuracy.