机械振动源的联合块对角化盲分离算法

基于QR分解建立一种新的非正交联合块对角化（joint block diagonalization,JBD）（QRJBD算法）的振动源盲分离方法.该方法具有对目标矩阵的限制少、复杂度低、易于收敛到全局最优解等优点.结合双层加肋圆柱壳体结构的振动信号分离试验,从算法的收敛性、目标矩阵的数目、随机噪声水平、子块矩阵的维数等方面对QRJBD方法的性能进行了研究.由此,选取合理的算法参数,实现振动源信号的盲源分离,且分离精度和时间均优于现存常用的方法,充分说明新算法在振源分离中既保持了效率又提高了分析的准确性.

Blind Separation of Coupling Mechanical Vibration Sources Based on Joint Block Diagonalization

One of the difficulty arises from the fact that a mixture of vibrations is most often of the convolutive type,which are a much more difficult to tackle.To improve existing BSS algorithms’ flaws,we present a class of simple Jacobi-type algorithms for non-orthogonal matrix JBD based on the QR factorizations,which have the merits of simplicity,effectiveness,and computational efficiency.A series of comparisons of the proposed algorithms with existing algorithms applied widely.Finally,the new JBD algorithms for practical vibration sources separation are proved and studied.The results demonstrate the robustness and performance improvement of the proposed algorithms,which can also effectively separate vibration sources.