邻域自适应选取及权值平均的投影去噪方法

针对常规局部投影方法在选取邻域点和还原时间序列两方面存在的不足,基于混沌时间序列的相空间重构理论,提出了一种基于相空间重构理论的改进的局部投影非线性去噪方法。该方法在邻域范围的选取和时间序列还原两方面进行了改进。用自适应的邻域选取方法代替了常规的固定邻域范围的方法,同时用加权平均代替直接平均还原时间序列。用改进后的方法对不同初始噪声水平下的Henon时间序列和Lorenz时间序列进行仿真,比较常规方法和改进方法去噪后的信噪比,仿真结果表明提出的改进方法能够更有效地去除包含在时间序列中的噪声成分,同时又能够较好地保留原系统中的混沌特性。

Improved local projection method by adaptive neighborhood selection and weighed average

A theory based on phase space reconstruction is introduced,and an improved local projection method on nonlinear noise reduction is presented to overcome the drawbacks of the conventional local projection noise reduction method.The improvement includes recovering time series with weighted methods and searching neighbors among aneighborhood adaptively.As the example of chaotic data,Henon time series and Lorenz time series are used to illustrate the methods.The result proves that the proposed method can reduce the noise more efficiently than conventional local projection methods,without destroying the chaotic characteristic of the system.