Bearing fault diagnosis using Radial Basis Function network and adaptive neuro fuzzy classifier

This paper is focused on the rolling element bearing fault diagnosis using radial basis function (RBF) network and adaptive neuro fuzzy classifier (ANFC). Defect considered for the study are inner race defect, outer race defect and ball defect. Vibration data for healthy bearing is used as baseline data. Multi scale permutation entropy (MPE) is applied to extract feature from the vibration signals. Then, attribute selection is done by best first search algorithm. A comparative study is carried out using ANFC and RBF network. Result shows that these methods has potential in diagnosis of incipient bearing defects and can be used for reliable fault diagnosis.