归一化RBF网络的时空混沌时间序列建模与应用

Spatiotemporal chaotic time series modeling using normalized RBF network and its application

本文提出将归一化RBF神经网络应用于时空混沌时间序列的建模与预测,并遗传算法训练该网络。通过利用该模型分别对参数可变的Lorenz混沌时间序列、耦合映象格子产生的时空混沌序列和真实的脑电信号进行建模和预测,得到较小的预测误差,证明了该模型对时空混沌时间序列有比较强的拟合能力和比较高的预测精度,有一定的工程应用价值。

In this paper, we propose a new method to model and predict the EEG signal based on the chaos dynamics, which is called normalized radial basis function network （NRBFNN）. This normalized RBF network is trained by genetic algorithm （GA）. The simulations with Parameter-Varying Lorenz time sequence,coupled map lattice （CML） time sequence and real EEG signal all evaluated the effectiveness of the proposed model. Therefore,to a certain extent,it performs the project application value.