期刊名称：International Journal of Multimedia and Ubiquitous Engineering
印刷版ISSN：1975-0080
出版年度：2015
卷号：10
期号：4
页码：231-242
DOI：10.14257/ijmue.2015.10.4.22
出版社：SERSC
摘要：Chaotic diagonal recurrent neural network is optimized and proposed to predict ship rolling motion. Aiming at the traditional arithmetic deductions of each weight value mediate do not give the specific sample time, it makes some lacks in this kind of deduction, the paper tries to optimize sampling time, carried out the derivation of the weight training and a convergence theorem of each weight learn algorithm based on Lyapunov function is given and proofed. Simulation results demonstrate the use of optimal sampling times increases the accuracy of all of the weight and algorithm convergence, to advance predicted precision so that forecast time of ten seconds efficiently. obviously better than using feed-forward BP neural network to predict.
关键词：chaotic diagonal recurrent neural network; ship rolling motion; momentum ; gradient learning algorithm; feed-forward BP neural network
