An Imbalanced Fault Diagnosis Method Based on TFFO and CNN for Rotating Machinery

Zhang, Long; Liu, Yangyuan; Zhou, Jianmin; Luo, Muxu; Pu, Shengxin; Yang, Xiaotong

doi:10.3390/s22228749

Cited by 10 publications

(8 citation statements)

References 52 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The theme of this Special Issue focuses on machine health monitoring and fault diagnosis techniques, especially intelligent fault diagnosis. This Special Issue highlights 18 articles that can be divided into four categories: condition monitoring [ 1 , 2 , 3 , 4 ], degradation process prediction [ 5 , 6 , 7 , 8 ], intelligent diagnostic algorithms [ 9 , 10 , 11 , 12 , 13 , 14 , 15 ], and sensor fault detection [ 16 , 17 , 18 ]. In addition to the traditional bearing vibration signals, the research objects include the electrode signals, blade vibration signals, diesel engine vibration signals, and bearing heat signals.…”

Section: Discussionmentioning

confidence: 99%

See 1 more Smart Citation

Editorial for Special Issue: Machine Health Monitoring and Fault Diagnosis Techniques

Sun

Shen

Wang

2023

Sensors

View full text Add to dashboard Cite

show abstract

Section: Discussionmentioning

confidence: 99%

“…In [ 9 ], a model for data augmentation was proposed. This study proposed a method for the unbalanced fault diagnosis of rotating machinery that combined time–frequency feature oversampling (TFFO) with a convolutional neural network (CNN).…”

Section: Fault Diagnosismentioning

confidence: 99%

Editorial for Special Issue: Machine Health Monitoring and Fault Diagnosis Techniques

Sun

Shen

Wang

2023

Sensors

View full text Add to dashboard Cite

show abstract

“…The decomposed low-frequency part is still relatively smooth and can continue to be wavelet decomposed, while the high-frequency part is increasingly detailed. The decomposed high-frequency part contains the fast changing conditions of the signal, which can capture the instantaneous behavioral characteristics of the signal [13,14]. As the number of wavelet decomposition layers increases, the variance and amplitude of the noise decreases, while the variance and amplitude of the useful signal increases, thus making the signal characteristics more pronounced.…”

Section: Wavelet Transformsmentioning

confidence: 99%

Research on the quality evaluation method of copper and aluminum diffusion bonding

Li,

Hou,

Yun

et al. 2024

Preprint

View full text Add to dashboard Cite

The application of copper-aluminum welded pole pieces is extensive in the new energy manufacturing sectors, such as automotive batteries. Diffusion bonding is one of the main ways of copper-aluminum connection, which is very important for the safe use of the components in terms of the quality of its connection. First, a water-immersion ultrasonic inspection platform was built to carry out inspection research on internal defects in copper-aluminum welding. In order to address the problem of overlapping signal features at the interface, the signal is decomposed and reconstructed by applying the wavelet decomposition method using the water-immersion focusing transmission method. The reconstructed signal is then decomposed in three layers, and the energy features of the last layer are extracted to form energy feature vectors for the qualitative analysis of different defect types. Secondly, different welding area samples were prepared and the peak power spectral density was extracted as an interface characterization parameter. Tensile tests were conducted on samples with varying weld areas, and the measured maximum pull-out force was correlated with the characteristic parameters of ultrasonic testing. The results demonstrate the effective capability of wavelet packet analysis in identifying un-welded, weak bond, and micro-porous defects at the diffusion welding interface. A higher peak value of power spectral density indicates superior welding quality. Establishing a reasonable threshold for peak power spectral density can serve as a reliable criterion for assessing the condition of copper and aluminum diffusion bonding.

show abstract

“…High-dimensional non-linear features can be transformed into low-dimensional features by using enough hidden layers, thus effectively capturing the hidden information in the data and achieving very complex learning functions [17]. Eren et al [18] used a convolutional neural network (CNN) classifier for bearing fault diagnoses; Zhang et al [19] proposed a fault diagnosis method that combines timefrequency feature oversampling (TFFO) with CNN; Yan [20] proposed a fault diagnosis model, MTF-ResNet, based on the Markov transition field and deep residual network; Liu et al [21] proposed a way of using RNN as an autoencoder for bearing fault diagnosis; Gao et al [22] optimized fault feature extraction by improving the WDCNN network; and Chen et al [23] proposed a novel Wide Residual Relation Network (WRRN) for RM intelligent fault diagnoses.…”

Section: Introductionmentioning

confidence: 99%

Method for Diagnosing Bearing Faults in Electromechanical Equipment Based on Improved Prototypical Networks

Wang

Shen

Wen-zhuo

et al. 2023

Sensors

View full text Add to dashboard Cite

Due to the complexity of electromechanical equipment and the difficulties in obtaining large-scale health monitoring datasets, as well as the long-tailed distribution of data, existing methods ignore certain characteristics of health monitoring data. In order to solve these problems, this paper proposes a method for the fault diagnosis of rolling bearings in electromechanical equipment based on an improved prototypical network—the weight prototypical networks (WPorNet). The main contributions of this paper are as follows: (1) the prototypical networks, which perform well on small-sample classification tasks, were improved by calculating the different levels of influence of support sample distributions in order to achieve the prototypical calculation. The change in sample influence was calculated using the Kullback–Leibler divergence of the sample distribution. The influence change in a specific sample can be measured by assessing how much the distribution changes in the absence of that sample; and (2) The Gramian Angular Field (GAF) algorithm was used to transform one-dimensional time series into two-dimensional vibration images, which greatly improved the application effect of the 2D convolutional neural network (CNN). Through experiments on MAFAULDA and CWRU bearing datasets, it was shown that this network effectively solves the shortcomings of a small number of valid samples and a long-tail distribution in health monitoring data, it enhances the dependency between the samples and the global data, it improves the model’s feature extraction ability, and it enhances the accuracy of model classification. Compared with the prototypical network, the improved network model increased the performance of the 2-way 10-shot, 2-way 20-shot, and 2-way 50-shot classification tasks by 5.23%, 5.74%, and 4.37%, respectively, and it increased the performance of the 4-way 10-shot, 4-way 20-shot, and 4-way 50-shot classification tasks by 12.02%, 10.47%, and 4.66%, respectively. Experimental results show that the improved prototypical network model has higher sample classification accuracy and stronger anti-interference ability compared with traditional small-sample classification models.

show abstract

An Imbalanced Fault Diagnosis Method Based on TFFO and CNN for Rotating Machinery

Cited by 10 publications

References 52 publications

Editorial for Special Issue: Machine Health Monitoring and Fault Diagnosis Techniques

Editorial for Special Issue: Machine Health Monitoring and Fault Diagnosis Techniques

Research on the quality evaluation method of copper and aluminum diffusion bonding

Method for Diagnosing Bearing Faults in Electromechanical Equipment Based on Improved Prototypical Networks

Contact Info

Product

Resources

About