基于小波多分辨分析的线性预测

Linear prediction based on wavelet multiresolution analysis

提出了一种对非平稳时间序列预测的新方法.通过小波多分辨分析把某些非平稳间序列分解若干层近似意义上的平稳时间序列,然后再用自回归模型对每层的单支重构信号进行预测,最后综合每层的预测值可得到原时间序列的预测值.仿真实验表明了该方法的优越性.

A prediction method of non-stationary time series is proposed. Some non-stationary time series can be decomposed into several approximate stationary time series by wavelet multiresolution analysis. The result of single branch reconstruction of every level is predicted ,and the prediction result of original time series with autoregressive model is obtained by synthesizing the prediction result of every level.Simulation indicates the advantages of this method.