基于混沌理论与径向基函数神经网络的混沌时间序列预测

Prediction of chaotic time series based on chaotic theory and radial basis function neural networks

根据混沌时间序列的特性 ,给出了将混沌理论与径向基函数神经网络相结合对其预测的方法 .首先在虚假邻域概念基础上 ,提出了可同时确定合适的嵌入维数与时间延迟的方法 ,从而可据此确定径向基函数神经网络的输入 ;然后 ,用径向基函数神经网络进行学习及预测 .最后 ,给出一个实例 .

This paper provides a method how to predict chaotic time series by connecting chaotic theory with radial basis function neural networks based on the properties of chaotic time series. First, based on the notion of the false nearest neighbors, a method is given to get correct embedding dimension and time lag simultaneously, thus, the inputs of radial basis function neural networks are ascertained. Then, people can use radial basis function neural networks to train and predict. Finally, a example is given.