Nonlinear Time Series Prediction Based on the Dynamic Characteristics Clustering Neural Network

Abstract

A novel nonlinear time series prediction method based on the dynamic clustering neural network is proposed. This method selects prediction samples which are similar with training samples according to the cluster analysis based on the dynamic characteristics of samples and then a new subset of the samples is obtained. All of the samples in this subset have the similar dynamic characteristics. By training with these samples, a model of BP neural network based on clustering is got and it is used in nonlinear time series prediction. Take the stress data of a large span bridge tower induced by strong typhoon as example. The results indicate the validity and the better prediction accuracy of this method in nonlinear time series prediction.