Incipient fault diagnosis based on ITD fractal dimension and fuzzy entropy for bearings

Aiming at the low signal to noise ratio and the limitation of single fault feature which difficultly achieves accurate diagnosis on a whole state area, an intelligent diagnosis method based on intrinsic time-scale decomposition (ITD), fractal dimension and fuzzy entropy is proposed. Firstly, the nonlinear and non-stationary vibration signals with plenty of noise are decomposed adaptively into series of proper rotation (PR) components. Secondly, the fractal dimension and fuzzy entropy of the main components containing the fault information are extracted as eigenvectors to express the fault information. Lastly, least squares support vector machine (LSSVM) is served as the approach of pattern recognition to identify fault categories. The incipient fault identification model of roller bearing with four work conditions is taken as an example. The analysis results show that, compared to single feature, the proposed method can be effectively applied to early fault intelligent diagnoses for roller bearings.