A Novel Deep Sparse Filtering Method for Intelligent Fault Diagnosis by Acoustic Signal Processing

Increased attention has been paid to research on intelligent fault diagnosis under acoustic signals. However, the signal-to-noise ratio of acoustic signals is much lower than vibration signals, which increases the difficulty of signal denoising and feature extraction. To solve the above defect, a no...

Full description

Saved in:
Bibliographic Details
Published inShock and vibration Vol. 2020; no. 2020; pp. 1 - 11
Main Authors Wang, Xiaoyu, Jia, Sixiang, Han, Baokun, Wang, Jinrui, Zhang, Guowei, He, Jingtao
Format Journal Article
LanguageEnglish
Published Cairo, Egypt Hindawi Publishing Corporation 2020
Hindawi
John Wiley & Sons, Inc
Hindawi Limited
Subjects
Acoustic noise
Acoustics
Algorithms
Back propagation
Computing time
Datasets
Diagnostic systems
Fast Fourier transformations
Fault diagnosis
Feature extraction
Filtration
Fourier transforms
Frequency domain analysis
Gear trains
Learning
Methods
Neural networks
Noise reduction
Rotating machinery
Signal processing
Signal to noise ratio
Sparsity
Teaching methods
Wavelet transforms
Online AccessGet full text

Cover

More Information
Summary:Increased attention has been paid to research on intelligent fault diagnosis under acoustic signals. However, the signal-to-noise ratio of acoustic signals is much lower than vibration signals, which increases the difficulty of signal denoising and feature extraction. To solve the above defect, a novel batch-normalized deep sparse filtering (DSF) method is proposed to diagnose the fault through the acoustic signals of rotating machinery. In the first stage, the collected acoustic signals are prenormalized to eliminate the adverse effects of singular samples, and then the normalized signal is transformed into frequency-domain signal through fast Fourier transform (FFT). In the second stage, the learned features are obtained by training batch-normalized DSF with frequency-domain signals, and then the features are fine-tuned by backpropagation (BP) algorithm. In the third stage, softmax regression is used as a classifier for heath condition recognition based on the fine-tuned features. Bearing and planetary gear datasets are used to validate the diagnostic performance of the proposed method. The results show that the proposed DSF model can extract more powerful features and less computing time than other traditional methods.
ISSN:1070-9622
1875-9203
DOI:10.1155/2020/8837047