摘要
采用基于非线性动力学系统的局部投影非线性降噪理论,研究了水声信号的非线性降噪问题,并对传统的局部投影算法进行了改进。采用自适应调整邻域大小的方法,改善了传统局部投影算法的邻域选择问题;利用局部投影算法中局部协方差矩阵存在特征值突变的现象,有效地解决了信号子空间和噪声子空间的划分难题。以Henon混沌时间序列为仿真模型,对传统的局部投影算法,以及改进后的算法的降噪性能进行了验证比较,结果表明,改进后的算法更加有效地抑制了噪声。
A noise reduction method for the underwater acoustic signal was discussed, based on the local-projection （LP） theory of the nonlinear dynamic system. A modified LP algorithm was proposed. Adaptively adjusted neighborhood was used to solve the problem of choosing neighborhood in the traditional LP algorithm. By means of the eigenvalue leap phenomenon that exists in the local covariance matrix in LP algorithm, it is effectively to overcome the difficulty in distinguishing the eigenvectors between signal space and noise space in the signal. Taking chaotic signals generated by Henon map for example, performances of traditional and modified LP algorithm were compared by the computer simulation. Compared with the traditional LP algorithm, the improved one reduces noise more effectively.
出处
《海洋通报》
CAS
CSCD
北大核心
2008年第1期17-21,共5页
Marine Science Bulletin
关键词
非线性降噪
局部投影
邻域选择
水声信号
nonlinear noise reduction
local-projection
neighborhood choice
underwater acoustic signal
