Article

Deep Residual Shrinkage Networks for Fault Diagnosis

Minghang ZhaoSchool of Naval Architecture and Ocean Engineering, Harbin Institute of Technology, Weihai, ChinaShisheng ZhongSchool of Naval Architecture and Ocean Engineering, Harbin Institute of Technology, Weihai, ChinaXuyun FuSchool of Naval Architecture and Ocean Engineering, Harbin Institute of Technology, Weihai, ChinaBaoping TangState Key Laboratory of Mechanical Transmission, Chongqing University, Chongqing, ChinaMichael PechtCenter for Advanced Life Cycle Engineering, University of Maryland, College Park, USA

2019en

ABI

Abstract

This article develops new deep learning methods, namely, deep residual shrinkage networks, to improve the feature learning ability from highly noised vibration signals and achieve a high fault diagnosing accuracy. Soft thresholding is inserted as nonlinear transformation layers into the deep architectures to eliminate unimportant features. Moreover, considering that it is generally challenging to set proper values for the thresholds, the developed deep residual shrinkage networks integrate a few specialized neural networks as trainable modules to automatically determine the thresholds, so that professional expertise on signal processing is not required. The efficacy of the developed methods is validated through experiments with various types of noise.

Identifiers

DOI: 10.1109/tii.2019.2943898

Citations and references

Cited by 20 references

Metrics — AkademScholar