摘要
本文通过三种方法分别研究了Lorenz模型和高斯白噪声，获得了混沌信号和噪声在这些方法下的不同表现特征。提出了根据关联维数（correlationdimension）判断降噪效果的方法，并进一步探讨了采用主分量分析（principalcomponentanalysis）确定嵌入维数（embeddingdimension）和去除混沌信号中噪声的可行性。最后得出结论：主分量分析用于确定嵌入维数和降低混沌信号中的噪声是不合适的。
In this paper, three methods have been used to characterize chaos signal and noise from Lorenz model and Gaussian White Noise. A new creterion has been proposed to quantify the error resulted from noise reduction ascording to correlation dimension. Next, the feasibility of principal component analysis to dermine embedding dimension and to reduce noise has been discussed. Finally a conclusion has been drawn that principal component analysis is not suitable for the determinition of embedding dimension and noise reduction.
出处
《信号处理》
CSCD
1997年第2期112-118,125,共8页
Journal of Signal Processing
基金
博士点基金
关键词
相空间重构
混沌信号
噪声
关联维数
FFT
phase space reconstruction
noise or chaos
fast Fourier transform
correlation dimension
principal component analysis, embedding dimension
noise reduction
