The application of NAH-based fault diagnosis method based on blocking feature extraction in coherent fault conditions

To further study the fault diagnosis method based on acoustic images, one near-field acoustical holography (NAH)-based fault diagnosis method is developed which firstly introduce the NAH technology and blocking feature extraction method into fault diagnosis. In allusion to the coherent fault conditions in which different machine components correspond to one and the same feature frequency and the coherent sound field is generated, one rib plate multi-excitation experiment is implemented. The scanning measurement technique is employed to sample the sound signals, and then the NAH algorithm is utilized to reconstruct the sound pressure distribution of sound sources for source recognition. Considering the physical meaning of the acoustic images, blocking feature extraction technique is applied to compose the eigenvectors. At last, the multiclass-support vector machine (SVM) is employed to train the feature vectors and diagnose the machine conditions. The experiments in laboratory demonstrate that the new diagnosis technology based on NAH technology is feasible and is one appropriate method comparing to acoustic-based diagnosis technique based on isolated point test in coherent fault conditions, and simultaneously widens the applications of NAH technique in the area of fault diagnosis.