摘要
探讨了基于相空间重构的局部线性映射算法在非线性时间序列降噪技术中的应用,并给出了算法中主要参数的选取方法.实验结果表明,该算法的降噪效果明显优于传统的线性信号滤波技术.并且针对多数实测数据的原始动态模型未知的特点,提出通过计算降噪前后时序信号的关联维数作为评判降噪效果的工具,克服了已有方法中无法计算该类时序信号降噪水平的缺点.
We analyze the local linear projective algorithm based phase space reconstruction in noise reduction techniques of nonlinear time series. How to choose the main parameters used in the method is also studied in this paper. We have found that new nonlinear method leads to much better results than traditional filtering techniques. Because in many experimental situations the noise-free data is unknown, it is difficult to investigate how much noise is removed from data. So we present that correlation dimension estimation from data before and after noise reduction provides a very good tool for filtering quality determination.
出处
《数学的实践与认识》
CSCD
北大核心
2007年第7期58-63,共6页
Mathematics in Practice and Theory
基金
上海市教委青年基金(04OC19)
关键词
非线性时间序列
相空间重构
噪声
关联维数
nonlinear time series
phase space reconstruction
noise reduction
correlation dimension
