A rub fault recognition method based on generative adversarial nets

Faced with the problem of valid data shortage data in practical. There's not enough data to train classifiers which can be satisfied to detect impact-rubbing faults in rotary machine. Bedsides, the large number of noises in working enviroment make the useful signal contaminated. Based on this problem, this paper proposes a rubbing fault recognition method based on a generative adversarial nets named deep convolution generative adversarial nets (DCGAN), which is based on a deep convolutional network frame with generation and discrimination models. The acquired signal is processed by time frequency analysis further to get spectrogram. The DCGAN can perform feature conversion and map it to the potential feature subspace to obtain more robust features. The results illustrate that the proposed method can achieve a much more excellent recognition effect. Thus, the proposed DCGAN model is an effective way to recognize impact-rubbing fault in the practical.

[1]  Li Zhao,et al.  Localization of Acoustic Emission Source Based on Chaotic Neural Networks , 2012 .

[2]  René Vinicio Sánchez,et al.  A Systematic Review of Fuzzy Formalisms for Bearing Fault Diagnosis , 2019, IEEE Transactions on Fuzzy Systems.

[3]  Yoshua Bengio,et al.  Generative Adversarial Nets , 2014, NIPS.

[4]  Georg Langs,et al.  Unsupervised Anomaly Detection with Generative Adversarial Networks to Guide Marker Discovery , 2017, IPMI.

[5]  Li Zhao,et al.  A new iterative near-field coherent subspace method for rub-impact fault localization using AE technique , 2017 .

[6]  Jung-Hoon Park,et al.  Acoustic emission characteristics of methacrylate-based composite and silorane-based composite during dental restoration according to a variety of C-factor , 2017 .

[7]  Li Zhao,et al.  Recognition of Acoustic Emission Signal based on the Algorithms of TDNN and GMM , 2014 .

[8]  Soumith Chintala,et al.  Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks , 2015, ICLR.

[9]  Jing Li,et al.  Mine Water Inrush Sources Online Discrimination Model Using Fluorescence Spectrum and CNN , 2018, IEEE Access.

[10]  S. Al-Dossary,et al.  Observations of changes in acoustic emission waveform for varying seeded defect sizes in a rolling element bearing , 2009 .

[11]  D Mba,et al.  Bearing defect diagnosis and acoustic emission , 2003 .

[12]  ZhiQiang Chen,et al.  Gearbox Fault Identification and Classification with Convolutional Neural Networks , 2015 .

[13]  Franck Vermet,et al.  Statistical Learning Methods , 2018 .

[14]  Hyunsoo Kim,et al.  Learning to Discover Cross-Domain Relations with Generative Adversarial Networks , 2017, ICML.

[15]  Sebastian Nowozin,et al.  f-GAN: Training Generative Neural Samplers using Variational Divergence Minimization , 2016, NIPS.

[16]  Léon Bottou,et al.  Wasserstein GAN , 2017, ArXiv.

[17]  Brendan J. Frey,et al.  Generating and designing DNA with deep generative models , 2017, ArXiv.

[18]  Amos J. Storkey,et al.  Data Augmentation Generative Adversarial Networks , 2017, ICLR 2018.

[19]  Aidong Deng,et al.  Study on Location Algorithms of Beamforming based on MVDR , 2013 .

[20]  Qingguo Fei,et al.  Prediction of the transient energy response for complex vibro-acoustic systems , 2019, Journal of Mechanical Science and Technology.

[21]  Terrence J. Sejnowski,et al.  An Information-Maximization Approach to Blind Separation and Blind Deconvolution , 1995, Neural Computation.

[22]  Diego Cabrera,et al.  A comparison of fuzzy clustering algorithms for bearing fault diagnosis , 2018, J. Intell. Fuzzy Syst..