Gearbox fault diagnosis based on multi-fractal

The multi-fractal analysis methods of nonlinear theory was adopted to extract multi-fractal characteristics of nonstationary vibration signals of the gear box under a variety of complex fault states, and training a variety of neural network to the classification of the composite fault signal. The experimental results show that the multi-fractal characteristics of extraction has a good degree of differentiation, combined with a neural network is an effective method of gearbox vibration signal analysis.