一种基于经验模态分解的时间序列预测方法  

A time series prediction method based on emirical mode decomposition

作　　者：马宇红 强亚蓉 杨梅 MA Yu-hong;QIANG Ya-rong;YANG Mei(College of Mathematics and Statistics,Northwest Normal University,Lanzhou 730070,Gansu,China;Editorial Department of the University Journal,Northwest Normal University,Lanzhou 730070,Gansu,China)

机构地区：[1]西北师范大学数学与统计学院,甘肃兰州730070 [2]西北师范大学学报编辑部,甘肃兰州730070

出　　处：《西北师范大学学报（自然科学版）》2020年第1期27-34,共8页Journal of Northwest Normal University(Natural Science)

基　　金：国家自然科学基金资助项目(51368055)

摘　　要：针对来源于实际问题的时间序列非线性、非平稳、多尺度复合的特点建立了一种基于经验模态分解(EMD)的ARIMA时间序列预测模型,即EMD-ARIMA模型.首先,借助经验模态分解将时间序列分解为多个不同时间尺度的内在模函数和一个趋势项,并确定每个内在模函数的季节性趋势;其次,对每个内在模函数使用季节性ARIMA模型进行预测,对趋势项使用趋势移动平均模型进行预测;最后,将所有内在模函数和趋势项的预测结果进行复合得到原时间序列的预测结果.数值实验结果表明,EMD-ARIMA方法能够揭示真实时间序列内在的多尺度复合特征和季节性变化规律;与经典的ARIMA模型和人工神经网络(ANN)模型相比,EMD-ARIMA模型明显提高了预测精度,因而是一种可靠的非线性、非平稳时间序列预测方法.A novel time series prediction model based on empirical mode decomposition(EMD)and ARIMA model is established according to the characteristics of nonlinear,nonstationary and multiscale composite for a large amount of time series existing in practical problems.Firstly,a time series is decomposed into several intrinsic mode functions(IMF)with different time scales and a trend term based on EMD method,and the periodic of each IMF is determined.Secondly,seasonal autoregressive integrated moving averaging(ARIMA)model is used to predict every IMF,while trend moving average(TMA)model is used to predict the trend term.Finally,the prediction results of all of sub-time series are combined to obtain final prediction results of the original time series.The experimental results show that EMD-ARIMA method not only reveals the characteristics of intrinsic multi-scale composite and laws of seasonal variation of actual time series,but also significantly improves the prediction accuracy of time series compared with the classical ARIMA model and artificial neural network(ANN)model,so it is a reliable prediction method for nonlinear and nonstationary time series.

关 键 词：时间序列 检验模态分解 内在模函数 季节性ARIMA模型 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程]