Journal of Applied Sciences ›› 2000, Vol. 18 ›› Issue (3): 208-210.

• Articles • Previous Articles     Next Articles

The Application of Wavelet in the Approximation of Non-linear Time Series

XIE Mei-ping1, ZHAO Xi-ren1, ZHUANG Xiu-long2   

  1. 1. Department if Automatic Control, Harbin Engineering University, Harbin 150001, China;
    2. Harbin Aeriadynamic Research Institute, Harbin 150001, China
  • Received:1999-09-21 Revised:2000-01-05 Online:2000-09-30 Published:2000-09-30

Abstract: The approximation properties of wavelet networks which is applied to non-linear time series is studied in this paper. Moreover, compared with projection pursuit learning networks and improved back propagation learning, wavelet networks have good fitness and forecasting properties.

Key words: wavelet analysis, neural network, non-linear time series model

CLC Number: