摘要
提出了一种基于小波分析理论的残差GM(1,1)-ARIMA组合预测方法.利用小波多尺度分解和重构思想分解非平稳时间序列,并对信号进行拟合,再将各模型预测值叠加,得到原始时间序列的预测值.仿真实例验证了本文方法的有效性.
A residual GM（1,1）-ARIMA combination forecasting method based on wavelet analysis was proposed.Using the non-stationary time series decomposed by the wavelet multis-cale decomposition and reconstruction,the predicted value of the original time series was obtained by fitting the signals and superposing all model forecasting values.Experiment results showed that this method is reasonable and achieves good prediction accuracy.
出处
《鲁东大学学报（自然科学版）》
2011年第4期305-309,341,共6页
Journal of Ludong University:Natural Science Edition
基金
教育部新世纪优秀人才支持计划(NCET-07-0404)
国家自然科学基金(40871208)
长江学者和创新团队发展计划资助(IRT0966)
