Extracting useful higher order features for condition monitoring using artificial neural networks

Author

Murray, A. ; Penman, J.

Author_Institution

Dept. of Eng., Aberdeen Univ., UK

Volume

45

Issue

11

fYear

1997

fDate

11/1/1997 12:00:00 AM

Firstpage

2821

Lastpage

2828

Abstract

Vibration data from an induction machine is employed to investigate higher order properties associated with electrical machine faults. Three fault conditions are investigated together with all possible permutations. By considering combinations of faults, interesting higher order properties are identified and presented, ultimately resulting in improved ANN diagnoses of faults

Keywords

acoustic signal processing; asynchronous machines; diagnostic expert systems; dynamic testing; electric machine analysis computing; fault diagnosis; feature extraction; higher order statistics; machine theory; neural nets; spectral analysis; transient analysis; vibration measurement; artificial neural networks; electrical machine faults; fault condition monitoring; fault diagnoses; higher order features extraction; higher order properties; induction machine; permutations; vibration data; Artificial neural networks; Condition monitoring; Data mining; Discrete Fourier transforms; Fault diagnosis; Frequency; Gaussian noise; Harmonic analysis; Higher order statistics; Induction machines;

fLanguage

English

Journal_Title

Signal Processing, IEEE Transactions on

Publisher

ieee

ISSN

1053-587X

Type

jour

DOI

10.1109/78.650108

Filename

650108