A Novel Method for Mechanical Fault Diagnosis of Underwater Pump Motors Based on Power Flow Theory

Due to difficulty in disposing of unsteady and nonlinear acoustic signals by conventional signal process methods, it remains a challenge to develop the noncontacting-based fault diagnosis techniques for underwater pump systems. Fortunately, the power flow theory (PFT), which has been proposed to ana...

Full description

Saved in:
Bibliographic Details
Published in:IEEE transactions on instrumentation and measurement Vol. 70; pp. 1 - 17
Main Authors: Jiang, Weixiong, Wang, Honghui, Liu, Guijie, Liu, Yonghong, Cai, Baoping, Li, Zhixiong
Format: Journal Article
Language:English
Published: New York IEEE 2021
The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
Subjects:
Acoustic noise
Acoustics
Back propagation
Back propagation networks
Convolution
Deep convolution processing feature (DCPM)
Empirical analysis
Fault diagnosis
Feature extraction
Flow theory
genetic algorithm-backpropagation neural network (GANN)
Genetic algorithms
Hilbert transformation
Load flow
Neural networks
Noise reduction
Power flow
power flow theory (PFT)
Signal processing
Sonar equipment
Underwater acoustics
underwater pump system
Underwater structures
Vibrations
ISSN:0018-9456, 1557-9662
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Due to difficulty in disposing of unsteady and nonlinear acoustic signals by conventional signal process methods, it remains a challenge to develop the noncontacting-based fault diagnosis techniques for underwater pump systems. Fortunately, the power flow theory (PFT), which has been proposed to analyze the fluid-solid interaction of underwater structures, provides great potential to process underwater acoustic signals. However, this potential has not been exploited in literature yet. In order to bridge this research gap, this article proposes a novel fault diagnosis method based on PFT, deep convolution processing method (DCPM), and genetic algorithm-backpropagation neural network (GANN). This method includes three steps: firstly, the complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN) is employed to decrease the proportion of the noise components in the original acoustic signals; and the radiated power flow is deduced from the denoised acoustics by PFT. Then, the Hilbert transform is conducted to obtain the Hilbert spectra of the radiated power flow signals and the deep convolution processing feature (DCPF) is extracted from the Hilbert spectra by DCPM. Lastly, the radiated power level (RPL) is calculated directly from the radiated power flow signals. The DCPF and RPL are input into a GANN for fault diagnosis. The effectiveness of the proposed method is validated using an underwater acoustic experiment. The results show that the diagnosis performance is competitive with the other five existing methods in terms of accuracy and efficiency.
Bibliography:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:0018-9456
1557-9662
DOI:10.1109/TIM.2020.3044300