Application of feature extraction based on fractal theory in fault diagnosis of bearing

Abstract

Fractal theory can be applied to state recognition and fault diagnosis of bearing for the nonlinear property of rotation machinery’s vibration signal. In this paper, a feature extraction method based on fractal theory is introduced and the fractal feature is extracted by computing the correlation dimension of vibration signals in different conditions. Correlation dimension can be determined by G-P algorithm and relevant parameters’ selection methods are discussed. C–C method is used to calculate the time delay of phase space reconstruction. The example of bearing shows that the correlation of bearing in fault condition is much higher than that in normal condition, which can help to recognize bearing’s state and discover bearing’s fault promptly.