Near field signal-subspace focusing model for rub-impact fault localization using acoustic emission data

In order to solve the fast-fading, coherent and multi-modes characteristics of wide-band acoustic emission(AE), a near field coherent subspace algorithm(N-CSM) for double AE sources in rub-impact fault localization is proposed. Firstly, according to the multi-modes property of material. the zero order mode of AE can be extracted using wavelet decomposition for interference removal and group velocity definition. Then, the two-side correlation transformation(TCT) is applied to automatically focus the covariance matrix on the focus frequency for correlative signals separation. Finally, the near field multiple signal classification algorithm(N-MUSIC) is implemented for double near field AE sources localizations. The results show that the proposed method has strong robustness, high location accuracy,low computational complexity and excellent performance on multiple coherent AE sources detection. Furthermore it can successfully avoid the initial estimation and optimal focus frequency calculation from classical coherent subspace algorithm(CSM). Resolution probability of double AE sources can also be improved 17% than the improved near field multiple signal classification(N-IMUSIC) algorithm. It is an effective way to detect the new rubbing fault position in rotary machine.