Adaptive Densely Connected Convolutional Auto-Encoder-Based Feature Learning of Gearbox Vibration Signals

Miao, Mengqi; Liu, Changhui; Yu, Jianbo

doi:10.1109/tim.2020.3039612

Cited by 19 publications

(7 citation statements)

References 27 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The DAE updates the weight and deviation of the encoding and decoding parts in each iteration process to achieve better data reconstruction. The gradient descent algorithm is used to update the corresponding weight and deviation to make the error between the original data and the reconstructed data as small as possible [17].…”

Section: Denoising Autoencoder(dae)mentioning

confidence: 99%

Research on pipeline inspection gauges localization based on residual denoising autoencoder

Guo,

Song,

Zeng

et al. 2023

Eng. Res. Express

View full text Add to dashboard Cite

To improve pipeline transportation efficiency, long-distance pipelines need to be cleaned regularly. Pipeline inspection gauges (PIGs) are widely used in pigging operations. The traditional PIG localization method generally has a large error and cannot achieve continuous tracking. To solve the problem of PIG real-time tracking and localization, this paper proposes a method of PIG localization for oil pipelines based on a denoising autoencoder(DAE). The attention deconvolution residual block (ADR) is introduced in the decoder part of the DAE, so that this model can learn and use global information to selectively emphasize key features, suppress useless features, and improve the noise removal performance of the model. After denoising, the negative pressure wave(NPW) signal is normalized, the time delay is extracted, and the PIG is located using the NPW localization formula. Experiments show that this method can effectively achieve real-time PIG localization. Compared with the EMD and DAE positioning methods, the relative positioning accuracy is increased by 3.5 times and 8.4 times, respectively.

show abstract

Section: Denoising Autoencoder(dae)mentioning

confidence: 99%

Research on pipeline inspection gauges localization based on residual denoising autoencoder

Guo,

Song,

Zeng

et al. 2023

Eng. Res. Express

View full text Add to dashboard Cite

show abstract

“…In recent years, the convolutional neural network (CNN), generative adversarial network (GAN), recurrent neural network (RNN), and other deep neural networks have been widely used in natural image denoising (Vo et al, 2021), medical electrocardiogram (ECG) signal denoising (Chiang et al, 2019), and audio signal denoising (Doumanidis et al, 2021). In particular, in vibration signal denoising processing, CNN can be used to effectively filter out a large number of noises and to complete the abstract feature extraction excellently (Miao et al, 2021). Liu et al (2019) combined the denoising autoencoder (DAE) and CNN to form a one-dimensional denoising convolutional autoencoder (DCAE-1D) network and achieved good denoising effect without additional preprocessing.…”

Section: Introductionmentioning

confidence: 99%

LN-MRSCAE: A novel deep learning based denoising method for mechanical vibration signals

Yang

Wang

et al. 2023

Journal of Vibration and Control

View full text Add to dashboard Cite

Vibration signals are used to monitor the running state of mechanical equipment, but always suffer from a lot of noise in the acquisition process. In order to eliminate noise interference as much as possible, a multilevel residual convolution autoencoder network based noise learning method (LN-MRSCAE) is proposed in this paper. The LN-MRSCAE consists of a deep convolutional autoencoder network and multilevel residual, in which the learning noise module encodes and decodes the noise components, combining with a residual module to obtain the denoised signal. The multilevel residual structure is designed to address the gradient disappearance in network learning and improve the training speed. In specific, the noise learning replaces the traditional clean signal learning. Instead of learning the characteristics of the original clean signal, just learn the characteristics of the noise signal. The effectiveness of LN-MRSCAE network denoising is verified by classic analog signal and actual bearing vibration signals with different faults. The results show that the LN-MRSCAE model is superior to the latest models, and the noise component of the signal is obviously suppressed.

show abstract

“…The advent of Squeezed-Excitation Networks (SENet) [30] allows the model to adaptively select feature map channels, thereby reducing noise. Miao et al [31] proposed a new adaptive densely connected convolutional autoencoder (ADCAE) for feature extraction directly from one-dimensional vibration signals, which allows unsupervised learning. Plakias et al [32] used attention-dense connected CNNs to accomplish similar work.…”

Section: Introductionmentioning

confidence: 99%

SRMANet: Toward an Interpretable Neural Network with Multi-Attention Mechanism for Gearbox Fault Diagnosis

Liu

Huang

et al. 2022

Applied Sciences

View full text Add to dashboard Cite

Deep neural network (DNN), with the capacity for feature inference and nonlinear mapping, has demonstrated its effectiveness in end-to-end fault diagnosis. However, the intermediate learning process of the DNN architecture is invisible, making it an uninterpretable black-box model. In this paper, a stacked residual multi-attention network (SRMANet) is proposed as a means of feature extraction of vibration signals, and visualizing the model training process, designing Squeeze-excitation residual (SE-Res) blocks to obtain additive features with minimal redundancy and sparsity. This study recommends the use of the attention fusion unit to ensure the interpretability of the model and ultimately to obtain representative features. By feeding the output gradient of the attention layer back to the original signal, the key feature components in the time domain signal can be effectively captured. Finally, the interpretability, identification accuracy and adaptability of the model under different operating conditions are verified on 12 different fault tasks in the planetary gearbox.

show abstract

Adaptive Densely Connected Convolutional Auto-Encoder-Based Feature Learning of Gearbox Vibration Signals

Cited by 19 publications

References 27 publications

Research on pipeline inspection gauges localization based on residual denoising autoencoder

Research on pipeline inspection gauges localization based on residual denoising autoencoder

LN-MRSCAE: A novel deep learning based denoising method for mechanical vibration signals

SRMANet: Toward an Interpretable Neural Network with Multi-Attention Mechanism for Gearbox Fault Diagnosis

Contact Info

Product

Resources

About