Affiliations: School of Reliability and Systems Engineering, Beihang
University, Beijing, China | Science and Technology Laboratory on Reliability and Environmental Engineering, Beijing, China
Note: [] Corresponding author: Chen Lu, Ph.D., School of Reliability and
Systems Engineering, Beihang University, Beijing 100191, China. Tel.: +86 10
8231 6573; E-mail: [email protected]
Abstract: Vibration signals extracted from rotating parts of machinery carry a
lot of useful information about the condition of operating machine. Due to the
strong non-linear, complex and non-stationary characteristics of vibration
signals from working bearings, an accurate and reliable health assessment
method for bearing is necessary. This paper proposes to utilize the selected
chaotic characteristics of vibration signal for health assessment of a bearing
by using self-organizing map (SOM). Both Grassberger-Procaccia algorithm and
Takens' theory are employed to calculate the characteristic vector which
includes three chaotic characteristics, such as correlation dimension, largest
Lyapunov exponent and Kolmogorov entropy. After that, SOM is used to map the
three corresponding characteristics into a confidence value (CV) which
represents the health state of the bearing. Finally, a case study based on
vibration datasets of a group of testing bearings was conducted to demonstrate
that the proposed method can reliably assess the health state of bearing.
Keywords: Health assessment, rolling bearing, feature extraction, chaotic characteristics, self-organizing map
