摘要
根据相空间重构理论,提出了一种基于遗传小波神经网络(GA-WNN)的混沌时间序列预测方法。根据takens理论,计算出相空间重构所需延迟时间和嵌入维数。采用小波神经网络的构造和算法,将遗传算法用于网络参数优化,为混沌时间序列预测提供可靠依据。为验证模型的可靠性,使用IPIX雷达数据进行多步预测,仿真结果表明这个确定性的模型可以根据海杂波已知数据预测未来值的变化。与传统神经网络预测相比,遗传小波神经网络预测方法的拟合精度和预测精度更好。
A method for chaotic time series prediction based on genetic algorithm wavelet neural network （GA-WNN） is discussed by analyzing the theory Of phase space reconstruction. According to the theory of Takens,calculate the phase space reconstruction for the time delay and embedding dimension. This paper presents the structure and algorithm of wavelet neural network and genetic algorithm, provide a reliable basis for the prediction of chaotic time series. In order to validate the reliability model,using IPIX radar datafor prediction,and results show the model of sea clutter can be based on known data predict future changes in value. Compared with the traditional neural network prediction, Genetic wavelet neural network prediction method better can get the fitting precision and prediction accuracy.
出处
《电子设计工程》
2015年第18期34-37,共4页
Electronic Design Engineering
