TITLE:
Improved Bearing Fault Diagnosis by Feature Extraction Based on GLCM, Fusion of Selection Methods, and Multiclass-Naïve Bayes Classification
AUTHORS:
Mireille Pouyap, Laurent Bitjoka, Etienne Mfoumou, Denis Toko
KEYWORDS:
GLCM, PCA, SFE, Naïve Bayes, Relevant Features
JOURNAL NAME:
Journal of Signal and Information Processing,
Vol.12 No.4,
October
18,
2021
ABSTRACT: The
presence of bearing faults reduces the efficiency of rotating machines and thus
increases energy consumption or even the total stoppage of the machine. It becomes essential to correctly diagnose the
fault caused by the bearing. Hence the importance of determining an
effective features extraction method that best describes the fault. The vision
of this paper is to merge the features selection methods in order to define the
most relevant featuresin the texture of the
vibration signal images. In this study, the Gray Level Co-occurrence Matrix
(GLCM) in texture analysis is applied on the vibration signal represented in
images. Features selection based on the merge of PCA (Principal component Analysis) method
and SFE (Sequential Features Extraction) method is done to obtain the most relevant features. The multiclass-Na?ve
Bayesclassifier is used to test the proposed approach. The success rate
of this classification is 98.27%. The relevant features obtained give promising
results and are more efficient than the methods observed in the literature.