基于振荡序列的GM(1,1)幂模型建模研究

摘要

对于数据变化并不是呈单调趋势,变化无规律的振荡序列,建模难度较大,预测效果不太理想.若采用时间跨度较大的数据进行建模,数据变化较大,其预测精度不高.采用时间间隔较小的数据建模,则数据的统计特征不能充分反映.为尽量保证建模预测的可靠性,利用灰色系统建模理论建立GM(1,1)幂模型,该模型体现了灰色系统的能量特征,充分利用数据特征,采用信息覆盖思想设定幂指数的白化公式,并给出GM(1,1)幂模型参数求解方法,较好地解决了模型参数计算的问题,拓展了GM(1,1)模型的使用范围.实证表明,GM(1,1)幂模型与GM(1,1)模型相比有效提高了模型的预测精度.%For the data change is not a monotonic trend and is the irregular oscillation sequence,the modeling is difficult,the prediction effect is not ideal. If the data with larger time span are used to model,the data change greatly and the prediction accuracy is not high. Data modeling with smaller time interval,the statistical characteristics of the data can not be fully reflected. To try to ensure the reliability of the modeling and prediction ,by using grey system modeling theory GM(1,1)power model is established. The model reflects the energy characteristics of grey system,makes use of the data features,sets power index whitening formula by using the information covered ideology,and gives the GM(1,1)power model parameter solving method,solves the problem of model parameter calculation,expands the use range of GM(1,1)model. Empirical results show that GM(1,1)power model compared with GM(1,1)model effectively improve the prediction accuracy of the model.