Nonlinear Time Series Prediction Using High Precision Neural Network

Abstract:

Article Preview

A new type of high precision back propagation (BP) neural network model was proposed and applied to nonlinear time series for improving its prediction accuracy. In order to optimize the neural network structure, it uses the correlation analysis to select the number of input node for BP neural network at first. Second, it uses grey clustering method to select the initial number of hidden node for BP neural network, then using the grey correlation analysis method to analyze the correlation degree between hidden node output and network output and according to the size of correlation degree to delete the redundant hidden nodes. Meanwhile, in order to improve model prediction accuracy, it increases the direct connection between the input layer and output layer. Finally, prediction results show that the proposed model has good prediction capability.

Info:

Periodical:

Edited by:

Zhixiang Hou

Pages:

745-748

DOI:

10.4028/www.scientific.net/AMM.48-49.745

Citation:

J. H. Zhou et al., "Nonlinear Time Series Prediction Using High Precision Neural Network", Applied Mechanics and Materials, Vols. 48-49, pp. 745-748, 2011

Online since:

February 2011

Export:

Price:

$35.00

In order to see related information, you need to Login.

In order to see related information, you need to Login.