Acoustic Fault Diagnosis of Rotor Bearing System

Early diagnosis of failures can prevent financial losses and industry downtime. In this article, the author proposes an early fault technique for rotor-bearing faults. The proposed is based on recognition sound signals. measured analyzed three states system: system under normal operating conditions, with faulty bearings, rotor friction. original feature extraction method described, namely, 1/3 doubling (a selecting amplitude frequency ratio that a multiple 30% maximum amplitude). This used to form vectors. A classification obtained vectors was performed by KNN (K-nearest neighbor classifier), SVM (support vector machine), decision tree. also compared Fourier synchrosqueezed transform. experimental results show diagnose faults systems simply quickly be protect safe operation mechanical equipment.