叶端定时中多重信号分类法的滤波特性研究

To reveal the reason why multiple signal classification (MUSIC) can filter out the synchronous frequency components the consistency between signal subspace of MUSIC and principal component of principal component analysis is theoretically investigated. One sinusoidal frequency component case is discussed to verify that the signal subspace of asynchronous frequency component and synchronous frequency component can be visually represented by a 2D cartesian coordinate and an axis respectively. Analyzing the position of the steering vector in the signal subspace, it is concluded that the artificial steering vector is outside the signal subspace of the synchronous frequency component, which leads to the filtering ability. The simulation and experiment results show that:synchronous frequency components and asynchronous frequency components coexist in the non-resonance region of blade vibration signal; MUSIC can eliminate aliasing andfilter out the synchronous frequency components; the filtering ability can be explained by the relationship between the steering vector and the subspace coordinate.