一种新的相关机械振源盲分离方法

Blind source separation of correlated vibration sources

针对传统盲源分离算法在机械振源不满足统计独立特性时,无法有效分离出振源信号的问题,提出了基于信号稀疏特性的相关机械源盲分离方法。盲源分离算法的关键在于准确地估计出混合矩阵。因此,首先提出了不相关源混合矩阵的估计方法;然后针对相关源,提出了有效剔除相关成分的方法,使得剩余信号可以按照不相关源进行处理。通过理论分析、仿真验证以及实测数据分析,验证了该方法的有效性。

The conventional blind source separation（BSS）method,independent component analysis（ICA）,may not separate correlated vibration sources.However,most vibration sources generated in different structures are correlated, they may cause failure of ICA.Here,a new BSS method capable of separating correlated vibration sources was proposed. It was divided into two steps.Firstly,it was needed to remove the correlated components,and then the reminder components were regarded as uncorrelated sources to be processed.Theoretical analysis,simulation and tests were employed to validate the effectiveness of the proposed method.