scholarly journals Rolling Bearing Incipient Fault Diagnosis Method Based on Improved Transfer Learning with Hybrid Feature Extraction

Sensors ◽  
2021 ◽  
Vol 21 (23) ◽  
pp. 7894
Author(s):  
Zhengni Yang ◽  
Rui Yang ◽  
Mengjie Huang
Keyword(s):  
Feature Extraction ◽  
Fault Diagnosis ◽  
Transfer Learning ◽  
Data Distribution ◽  
Rolling Bearing ◽  
Data Driven ◽  
Bearing Fault ◽  
Bearing Fault Diagnosis ◽  
Incipient Fault ◽  
Incipient Fault Diagnosis

Data-driven based rolling bearing fault diagnosis has been widely investigated in recent years. However, in real-world industry scenarios, the collected labeled samples are normally in a different data distribution. Moreover, the features of bearing fault in the early stages are extremely inconspicuous. Due to the above mentioned problems, it is difficult to diagnose the incipient fault under different scenarios by adopting the conventional data-driven methods. Therefore, in this paper a new unsupervised rolling bearing incipient fault diagnosis approach based on transfer learning is proposed, with a novel feature extraction method based on a statistical algorithm, wavelet scattering network, and a stacked auto-encoder network. Then, the geodesic flow kernel algorithm is adopted to align the feature vectors on the Grassmann manifold, and the k-nearest neighbor classifier is used for fault classification. The experiment is conducted based on two bearing datasets, the bearing fault dataset of Case Western Reserve University and the bearing fault dataset of Xi’an Jiaotong University. The experiment results illustrate the effectiveness of the proposed approach on solving the different data distribution and incipient bearing fault diagnosis issues.

Feature extraction for rolling bearing fault diagnosis by electrostatic monitoring sensors

2014 ◽  
Vol 229 (10) ◽  
pp. 1887-1903 ◽  
Author(s):  
Ying Zhang ◽  
Hongfu Zuo ◽  
Fang Bai
Keyword(s):  
Feature Extraction ◽  
Fault Diagnosis ◽  
Hankel Matrix ◽  
Rolling Bearing ◽  
Difference Spectrum ◽  
Permutation Entropy ◽  
Feature Extraction Method ◽  
Bearing Fault ◽  
Quantitative Indicator ◽  
Bearing Fault Diagnosis

There are mainly two problems with the current feature extraction methods used in the electrostatic monitoring of rolling bearings, which affect their abilities to identify early faults: (1) since noises are mixed in the electrostatic signals, it is difficult to extract weak early fault features; (2) traditional time and frequency domain features have limited ability to provide a quantitative indicator of degradation state. With regard to these two problems, a new feature extraction method for rolling bearing fault diagnosis by electrostatic monitoring sensors is proposed in this paper. First, the spectrum interpolation is adopted to suppress the power-frequency interference in the electrostatic signal. Then the resultant signal is used to construct Hankel matrix, the number of useful components is automatically selected based on the difference spectrum of singular values, after that the signal is reconstructed to remove background noises and random pulses. Finally, the permutation entropy of the denoised signal is calculated and smoothed using the exponential weighted moving average method, which is used to be a quantitative indicator of bearing performance state. The simulation and experimental results show that the proposed method can effectively remove noises and significantly bring forward the time when early faults are detected.

scholarly journals Deep Transfer Learning Method Based on 1D-CNN for Bearing Fault Diagnosis

2021 ◽  
Vol 2021 ◽  
pp. 1-16
Author(s):  
Jun He ◽  
Xiang Li ◽  
Yong Chen ◽  
Danfeng Chen ◽  
Jing Guo ◽  
...  
Keyword(s):  
Fault Diagnosis ◽  
Transfer Learning ◽  
Vibration Signal ◽  
Rolling Bearing ◽  
Learning Method ◽  
Target Domain ◽  
Source Domain ◽  
Second Order Statistics ◽  
Bearing Fault ◽  
Bearing Fault Diagnosis

In mechanical fault diagnosis, it is impossible to collect massive labeled samples with the same distribution in real industry. Transfer learning, a promising method, is usually used to address the critical problem. However, as the number of samples increases, the interdomain distribution discrepancy measurement of the existing method has a higher computational complexity, which may make the generalization ability of the method worse. To solve the problem, we propose a deep transfer learning method based on 1D-CNN for rolling bearing fault diagnosis. First, 1-dimension convolutional neural network (1D-CNN), as the basic framework, is used to extract features from vibration signal. The CORrelation ALignment (CORAL) is employed to minimize marginal distribution discrepancy between the source domain and target domain. Then, the cross-entropy loss function and Adam optimizer are used to minimize the classification errors and the second-order statistics of feature distance between the source domain and target domain, respectively. Finally, based on the bearing datasets of Case Western Reserve University and Jiangnan University, seven transfer fault diagnosis comparison experiments are carried out. The results show that our method has better performance.

scholarly journals Research on rolling bearing fault diagnosis based on multi-dimensional feature extraction and evidence fusion theory

2019 ◽  
Vol 6 (2) ◽  
pp. 181488 ◽  
Author(s):  
Jingchao Li ◽  
Yulong Ying ◽  
Yuan Ren ◽  
Siyu Xu ◽  
Dongyuan Bi ◽  
...  
Keyword(s):  
Pattern Recognition ◽  
Feature Extraction ◽  
Fault Diagnosis ◽  
Real Time ◽  
Rolling Bearing ◽  
Diagnostic Methods ◽  
Bearing Fault ◽  
Feature Vectors ◽  
Bearing Fault Diagnosis ◽  
Fusion Theory

Rolling bearing failure is the main cause of failure of rotating machinery, and leads to huge economic losses. The demand of the technique on rolling bearing fault diagnosis in industrial applications is increasing. With the development of artificial intelligence, the procedure of rolling bearing fault diagnosis is more and more treated as a procedure of pattern recognition, and its effectiveness and reliability mainly depend on the selection of dominant characteristic vector of the fault features. In this paper, a novel diagnostic framework for rolling bearing faults based on multi-dimensional feature extraction and evidence fusion theory is proposed to fulfil the requirements for effective assessment of different fault types and severities with real-time computational performance. Firstly, a multi-dimensional feature extraction strategy on the basis of entropy characteristics, Holder coefficient characteristics and improved generalized box-counting dimension characteristics is executed for extracting health status feature vectors from vibration signals. And, secondly, a grey relation algorithm is used to calculate the basic belief assignments (BBAs) using the extracted feature vectors, and lastly, the BBAs are fused through the Yager algorithm for achieving bearing fault pattern recognition. The related experimental study has illustrated the proposed method can effectively and efficiently recognize various fault types and severities in comparison with the existing intelligent diagnostic methods based on a small number of training samples with good real-time performance, and may be used for online assessment.

Feature Extraction of Rolling Bearing Fault Diagnosis

2012 ◽  
Vol 190-191 ◽  
pp. 993-997
Author(s):  
Li Jie Sun ◽  
Li Zhang ◽  
Yong Bo Yang ◽  
Da Bo Zhang ◽  
Li Chun Wu
Keyword(s):  
Feature Extraction ◽  
Fault Diagnosis ◽  
Attribute Reduction ◽  
Vibration Signal ◽  
Rolling Bearing ◽  
Mechanical Equipment ◽  
Characteristic Parameters ◽  
Bearing Fault ◽  
Bearing Fault Diagnosis ◽  
Experimental Trials

Mechanical equipment fault diagnosis occupies an important position in the industrial production, and feature extraction plays an important role in fault diagnosis. This paper analyzes various methods of feature extraction in rolling bearing fault diagnosis and classifies them into two big categories, which are methods of depending on empirical rules and experimental trials and using objective methods for screening. The former includes five methods: frequency as the characteristic parameters, multi-sensor information fusion method, rough set attribute reduction method, "zoom" method and vibration signal as the characteristic parameters. The latter includes two methods: sensitivity extraction and data mining methods to select attributes. Currently, selection methods of feature parameters depend heavily on empirical rules and experimental trials, thus extraction results are be subjected to restriction from subjective level, feature extraction in the future will develop toward objective screening direction.

Sign in / Sign up

Export Citation Format

Share Document