Research on Fault Diagnosis of Rolling Bearings Based on Fractal Dimension

Abstract

The key to fault diagnosis of mechanical equipment is the extraction of fault features. Due to the nonlinearity and non-stationarity of bearing vibration signals, fractal dimension, as a means of characterizing nonlinear features, can effectively describe the nonlinear characteristics of things. This article first introduces the single-fractal box dimension, correlation dimension and multi-fractal de-trend wave method algorithms. Then, with the help of FFT analysis and EMD decomposition of the rolling bearing normal state, inner ring failure, outer ring failure, and rolling element failure data collected by Western Reserve University, the IMF components of different characteristic frequency bands are obtained, and they are combined with the box dimension fractal method and The correlation dimension method is combined for bearing fault analysis. Finally, the multi-fractal de-trend wave method is used to analyze the vibration signal of the bearing in different states, and the random forest algorithm are used to verify and compare with the single fractal dimension. The results show that the combination of single fractal box dimension and correlation dimension combined with IMF can effectively perform fault analysis, but there are limitations in describing nonlinear characteristics, and the fault characteristic parameters obtained by multifractal detrending are more significant.