摘要： 相关源不满足独立分量分析关于源的统计独立性假设，标准的独立分量分析方法无法准确分离相关机械振源信号。在相关振源源信号的部分频带满足统计独立的假设前提下，本文提出了一种基于小波包分解的相关机械源盲源分离方法。该方法将观测信号用小波包分解成子带观测信号，根据互信息标准选择相关性较小的若干子带观测信号重构观测信号。通过重构的观测信号的独立分量分析估计分离矩阵，然后用该矩阵分离原始观测信号从而实现相关机械振源信号的分离。仿真试验验证了该方法的有效性。

Abstract: Standard independent component analysis is unable to separate statistically correlated machine vibration sources because the assumption of statistical independent can not be satisfied. Assumed some sub-component of correlated machine vibration sources are independent ,a novel blind source separation of correlated vibration sources based on wavelet packat decompostion was proposed. Firstly, subband observed signals was obtaied by decomposing orignal observed signals with wavelet packet. Secondly, the new observed signals was reconstructed by some subband observed signals with less dependent according to the mutual information creterion. Last, separation matrix was estimated through the new observed signals with independent component analysis; then, the correlated machine vibration sources was separated by estimated separation matrix. Simulations testfied the validity of the proposed method.