2020
DOI: 10.3390/app10186385
|View full text |Cite
|
Sign up to set email alerts
|

Accurate Bearing Fault Diagnosis under Variable Shaft Speed using Convolutional Neural Networks and Vibration Spectrogram

Abstract: Predicting bearing faults is an essential task in machine health monitoring because bearings are vital components of rotary machines, especially heavy motor machines. Moreover, indicating the degradation level of bearings will help factories plan maintenance schedules. With advancements in the extraction of useful information from vibration signals, diagnosis of motor failures by maintenance engineers can be gradually replaced by an automatic detection process. Especially, state-of-the-art methods using deep l… Show more

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
4
1

Citation Types

0
38
0

Year Published

2020
2020
2024
2024

Publication Types

Select...
7
2

Relationship

1
8

Authors

Journals

citations
Cited by 64 publications
(38 citation statements)
references
References 14 publications
0
38
0
Order By: Relevance
“…SNR dB = 10 log 10 P signal P noise (5) The experimental results in noisy conditions are shown in Table 4, which indicate that in lower SNR values, the accuracy of the proposed method becomes relatively worse, but still provides high accuracy. When the SNR values are positive, the average accuracy is more than 89.85% and 92.11% for Dataset 1 and Dataset 2, respectively.…”
Section: Compound Fault Diagnosis In Noisy Conditionsmentioning
confidence: 99%
See 1 more Smart Citation
“…SNR dB = 10 log 10 P signal P noise (5) The experimental results in noisy conditions are shown in Table 4, which indicate that in lower SNR values, the accuracy of the proposed method becomes relatively worse, but still provides high accuracy. When the SNR values are positive, the average accuracy is more than 89.85% and 92.11% for Dataset 1 and Dataset 2, respectively.…”
Section: Compound Fault Diagnosis In Noisy Conditionsmentioning
confidence: 99%
“…Therefore, many studies have focused on bearing FD using various types of signals acquired from sensors mounted on electric motors, especially vibration signals. In the last decade, with the advent of machine learning techniques, bearing FD methods using vibration signals have achieved high accuracy under specific conditions [5][6][7]. However, in realistic scenarios such as diagnosis of very early developed damage [3], diagnosis of very slow rotating machinery, and monitoring under high-vibration environments, acoustic…”
Section: Introductionmentioning
confidence: 99%
“…However, their work does not use other information available from signal data like skewness, kurtosis, impulse factor, RMS value [16]. Another work done by Pham et al proposes a method that converts the signal data into its spectrogram which is then fed to VGG16 for classification [17], [18]. In their paper, they used only four classes and achieved 98.8% accuracy.…”
Section: Introductionmentioning
confidence: 99%
“…With respect to the frequency domain, the decomposition of the signal by the Fourier transform [ 27 ], the Hilbert–Huang transform (based on EMD) [ 28 ], and the continuous wavelet transform [ 29 ] can be found. In addition, the analysis of the signal by the kurtogram application [ 30 ] or spectrogram [ 31 ] are used to detect changes in the frequency characteristics. In general, the decomposition of the vibration signals for detecting and diagnosing low-speed bearing damage are not comparable to the results obtained by their application to regular bearings [ 32 ].…”
Section: Introductionmentioning
confidence: 99%