Research on the Magnitude Time Series Prediction Based on Wavelet Neural Network

Abstract:

Based on the fundamental principles of the wavelet analysis combining with BP neural network, the paper can obtain the minimum embedding dimension and delay time. According to the chaos theory, the phase space of the magnitude time series can be reconstructed by Takens theorem. The paper uses wavelet neural network to train and test the nonlinear magnitude time series in the reconstructed phase space. The simulation results show that the predictive effect of the magnitude time series is remarkable and the predictive performance of single-step prediction is superior to that of multi-step prediction.

• Periodical: Applied Mechanics and Materials (Volumes 128-129)
• Pages: 233-236
• DOI: https://doi.org/10.4028/www.scientific.net/AMM.128-129.233
• Online since: October 2011
• Authors: Yan Lan Chen, Yi Chen, Qing Huang
• Keywords: Magnitude Prediction, Phase Space Reconstruction, Time-Series, Wavelet Neural Network (WNN)

References

[1] YANG Ju-yi, YI Yong-hong. Research on Earthquake Forecast [J]. Microelectronics and Computer, 2008, 25(10): 129-132.

[2] HAN Min. Prediction Theory and Methods of Chaotic Time Series [M]. Beijing: China WaterPower Press, 2007: 45-54.

[3] SHI Feng, etc. The Analysis of 30 Cases of MATLAB and Neural Network [M]. Beijing: Beijing University of Aeronautics & Astronautics Press, 2010: 208-217.

[4] SHI Yan, HAN Li-qun, LIAN Xiao-qin. The neural network design methods and instance analysis [M]. Beijing: Beijing University of Posts and Telecommunications Press, 2009: 23-27.

[5] LV Jin-hu, LU Jun-an, CHEN Shi-hua. Chaotic Time Series Analysis and Application [M]. Wuhan: Wuhan University Press, 2002: 58-63.