Mechanical Fault Identification Method Based on Vector Power Spectrum Coupled with Radial Basis Probabilistic Networks

A new mechanical fault identification method coupling vector power spectrum with radial basis probabilistic neural networks (RBPNN) is proposed in the paper. Vector power spectrum is used as eigenvectors, and radial basis probabilistic neural network (RBPNN) is used as a classifier in the new method. The method is used to identify the typical mechanical fault. The result shows that the new method is very effective to identify the fault diagnosis of rotating machinery, and has higher correct identification rate and faster training speed.