混沌序列的模糊神经网络预测

Using a Fuzzy Neural Network to Predict Chaotic Time Series

提出了一种针对混沌序列预测的T-S模糊神经网络。这种T-S模糊神经网络与传统的T-S模糊神经网络相比在不影响预测精度的前提下极大的减少了神经网络的节点数。同时利用基于混沌动态量的BP算法对神经网络进行学习,提高了收敛速度和预测精度。应用此T-S模糊神经网络和相应的BP学习算法,对Ma-ckey-Glass混沌时间序列进行了预测,与传统的的T-S模糊神经网络相比得到了更好的结果,验证了该方法的有效性。

In this paper the authors present a T-S neurofuzzy network to predict chaotic serial. The proposed neurofuzzy network has as precision as traditional neurofuzzy network, but it considerably reduce the number of point. At the same time the authors sue improved BP arithmetic and boost the time of constringency. Using the T-S eurofuzzy network and it's arithmetic to predict Mackey-Glass chaotic time serial, we receive better result and prove that the method that proposed in this paper is valid.