基于递推合成BP网络的多变量时间序列预测模型

提供了一种基于递推合成BP网络的非线性时间序列预测方法,并针对具体实例建立多变量时间序列模型.将其预测结果与灰色预测模型及常规BP网络的多变量时间序列预测模型的结果进行比较,其仿真实验结果表明该网络具有很强的学习特性和泛化能力,适合进行非线性时间序列建模及预测.

Multivariable Time Series Forecasting Model Based on Recurrent Composite BP Neural Networks and Its Application

This paper,presents a nonlinear time series forecasting model based on recurrent composite BP networks,It also establishes a multivariable time series model is aimed at concrete examples. Result of forecast is contrasted with that of the multivariable time series forecasting model based on grey theory and common BP networks,and its simulated result shows that the multivariable time series forecasting model based on recurrent composite BP networks has strong study feature and generalization ability and can be adapted to careate nonlinear time series model and forecast accurately.