基于EMD和神经网络的非线性时间序列预测方法

Prediction Method for Nonlinear Time Series Based on EMD and Neural Networks

现实生活中的时间序列,通常伴随着大量的噪声和高度的波动性。对于这些非线性时间序列,运用传统的统计和计量经济模型进行分析预测,预测结果往往不够理想。文章基于经验模态分解(EMD)和人工神经网络提出改进方法。主体思想是"先分再合":先用EMD方法分解非线性时间序列,得到一系列易于分析的独立的子系列,然后利用神经网络(FNN)对每一个子系列进行分析和预测,最后再用自适应线性神经网络(ALNN)整合并得出最终结果。结合具体房价时间序列实例,证实了这种方法的优势。

The time series in real life is usually accompanied by a lot of noise and high volatility.For these nonlinear time series,if traditional statistical and econometric models are used to make analyses and predictions,the results are often less than ideal.This paper proposes an improved method based on empirical mode decomposition(EMD)and artificial neural network.The main idea of the proposed method is"split first and then integrate":firstly,the EMD method is used to decompose the nonlinear time series so as to obtain a series of independent sub-series which are easy to analyze;then the feed-forward neural network(FNN)is used to analyze and predict each subseries;finally,the adaptive linear neural network(ALNN)is employed to integrate and obtain the final result.The paper also proves the advantages of the proposed method by combining with the time series of the concrete real estate price.