改进的带参考信号盲源分离算法

Improved Blind Source Separation Algorithm with Reference Signal

带参考信号的盲源分离算法在各个领域有着广泛的应用,但现有算法大都存在提取信号与源信号之间误差较大的问题,其中目标函数是影响误差的一个重要因素。因此针对目标函数,提出了一种改进带参考信号的盲源分离算法。该算法首先在标准对比函数中耦合含有先验信息的测量度函数,以此得到新的目标函数;然后引入松弛因子运用拉格朗日乘子法对目标函数进行优化,避免了不等式约束问题,有效地得到了最优的分离矩阵。仿真实验结果表明,相比现有算法,本文算法具有更小的误差;在滚动轴承故障诊断实验中也正确地提取了故障特征,验证了算法的有效性。

Blind source separation algorithms with reference signal are widely used in various fields,but most of the existing algorithms have the problem of large errors between the extracted signal and the source signal,in which the objective function is an important factor affecting the error.Therefore,aiming at the objective function,an improved blind source separation algorithm with reference signal was proposed.Firstly,the standard contrast function was coupled with the measurement function containing a priori information to obtain a new objective function.Then the relaxation factor was introduced and the Lagrange multiplier method was used to optimize the objective function,which avoided the inequality constraint problem and effectively obtained the optimal separation matrix.Simulation results show that the proposed algorithm has smaller errors compared with the existing algorithms.In the rolling bearing fault experiment,the fault features are correctly extracted,which verifies the effectiveness of the algorithm.