Electric Motor Fault Detection Using Fusion of Acoustic Signal Features

Abbas Ramezani | Saeed Seyedtabaii

---

URL :   http://research.shahed.ac.ir/WSR/WebPages/Report/PaperView.aspx?PaperID=158636
Date :  2021/09/11
Publish in :    ششمين کنفرانس ملي مهندسي مکانيک و هوافضا

Link :  http://later is added
Keywords :Fault Diagnosis, Acoustic Signal, Single phase motor, Fast Fourier Transform, Wavelet, ANN.

Abstract :

---

Non-destructive fault detection of industrial induction motors and their gearboxes is widely required. Faults may be shorted coils of the auxiliary or primary winding, broken rotor bar, and looseness between two gears of the gearbox. In this respect, statistical, spectrum, Shannon Entropy, and wavelet decomposition features of the signals are computed. The fusion of features is applied to enhance the classification performance, and the Principle component analysis (PCA) technique is employed to remove the redundancy among features and compact them. The test results indicate that the wavelet features are work relatively well. However, the best and reliable outcomes are obtained where the fusion of spectrum and wavelet features is utilized, leading to 100 accuracy.

http://later is added