论文部分内容阅读
针对传统欠定盲源分离方法难以有效处理具有时间延迟效应的发动机混叠振动信号分离问题,采用稀疏分量分析法研究了一种时间延迟混合信号的欠定盲源分离方法。该方法通过对混合信号的频域模值的线性聚类分析实现源数目及衰减矩阵的估计,并通过恢复混合信号的频域实部在混合空间的聚类性推断出时间延迟矩阵。实验结果表明,该方法在欠定条件下(传感器个数小于振动源个数)能够有效地将发动机混叠振动信号按照不同的激振源进行分离。
Aiming at the problem that the traditional underdetermined blind source separation method can not effectively deal with the problem of separation of aliasing vibration signals with time delay effect, a method of sparse component analysis for underdetermined blind source separation of time-delayed mixed signals is studied. The method estimates the number of sources and the attenuation matrix by linear clustering analysis of the frequency domain modal of the mixed signal, and deduces the time delay matrix by restoring the clustering of the real part of the mixed signal in the mixed space. Experimental results show that under the condition of underdetermined condition (the number of sensors is less than the number of vibration sources), this method can effectively separate the vibration signals of engine with different excitation sources.