Ball bearing is an important type of bearings. The radial acceleration of ball bearings has been measured for monitoring and diagnosis. Feature extraction is used to extract essential features from the experimental data. Three features, including peak amplitude of the frequency domain, percent power, and peak RMS, have been extracted from the radial acceleration of ball bearings. Then Sequential Forward Search Algorithm (SFS） was utilized for feature selection in order to effectively obtain the best vibration features. Adaptive Neuro Fuzzy Inference Systems (ANFIS) have been used. The selected features were the inputs to the neuro-fuzzy system. Whether there is a defect or not and what types of defects were the outputs of this system. Although there is no analytical relationship between the input and the output of the neuro-fuzzy system, this system still can establish the input/output relationship. In other words, this approach can most accurately, most quickly, and most reliably determine whether there is a defect or not and what types of defects, which is very important for preventive monitoring, diagnosis, and maintenance of ball bearings.

Date Published: 10 October 2013
PDF: 7 pages
Proc. SPIE 8916, Sixth International Symposium on Precision Mechanical Measurements, 89160B (10 October 2013); doi: 10.1117/12.2035880

Published in SPIE Proceedings Vol. 8916:
Sixth International Symposium on Precision Mechanical Measurements
Shenghua Ye; Yetai Fei, Editor(s)