Knowledge extraction from deep convolutional neural networks applied to cyclo-stationary time-series classification

Abstract Modelling complex processes from raw time series increases the necessity to build Deep Learning (DL) architectures that can manage this type of data structure. However, as DL models become deeper, larger and more diverse datasets are necessary and knowledge extraction will become more difficult. In an attempt to sidestep these issues, in this paper a methodology based on two main steps is presented, the first being to increase size and diversity of time-series datasets for training, and the second to retrieve knowledge from the obtained model. This methodology is compared with other approaches reported in the literature and is tested under two configuration setups of Condition-Based Maintenance problems: fault diagnosis of bearing, and fault severity assessment of a helical gearbox, obtaining not only a performance improvement in comparison, but also in retrieving knowledge about how the signals are being classified.

[1]  Nikola Kasabov,et al.  Evolving connectionist systems , 2002 .

[2]  Diego Cabrera,et al.  Automatic feature extraction of time-series applied to fault severity assessment of helical gearbox in stationary and non-stationary speed operation , 2017, Appl. Soft Comput..

[3]  Yaguo Lei,et al.  Fault detection of planetary gearboxes using new diagnostic parameters , 2012 .

[4]  Jason Lines,et al.  Classification of time series by shapelet transformation , 2013, Data Mining and Knowledge Discovery.

[5]  David,et al.  A comparative experimental study on the diagnostic and prognostic capabilities of acoustics emission, vibration and spectrometric oil analysis for spur gears , 2007 .

[6]  Nikola Kasabov,et al.  Springer Handbook of Bio-/Neuro-Informatics , 2013 .

[7]  Stefan Schliebs,et al.  Evolving spiking neural network—a survey , 2013, Evolving Systems.

[8]  Diego Cabrera,et al.  Fault diagnosis in spur gears based on genetic algorithm and random forest , 2016 .

[9]  Snjezana Soltic,et al.  Knowledge Extraction from Evolving Spiking Neural Networks with Rank Order Population Coding , 2010, Int. J. Neural Syst..

[10]  B. B. Seth,et al.  Analysis of repetitive mechanism signatures , 1980 .

[11]  José Valente de Oliveira,et al.  A Bayesian approach to consequent parameter estimation in probabilistic fuzzy systems and its application to bearing fault classification , 2017, Knowl. Based Syst..

[12]  Norman M. Abramson,et al.  A Generalization of the Sampling Theorem , 1960, Inf. Control..

[13]  Diego Cabrera,et al.  Fault diagnosis of spur gearbox based on random forest and wavelet packet decomposition , 2015 .

[14]  G. Arfken.  Mathematical Methods for Physicists , 1967 .

[15]  Diego Cabrera,et al.  A review on data-driven fault severity assessment in rolling bearings , 2018 .

[16]  Diego Cabrera,et al.  Gearbox fault diagnosis based on deep random forest fusion of acoustic and vibratory signals , 2016 .

[17]  Kunihiko Fukushima,et al.  Neocognitron: A self-organizing neural network model for a mechanism of pattern recognition unaffected by shift in position , 1980, Biological Cybernetics.

[18]  Yuvin Chinniah,et al.  Analysis and prevention of serious and fatal accidents related to moving parts of machinery , 2015 .

[19]  Carmelo J. A. Bastos Filho,et al.  Using a Support Vector Machine Based Decision Stage to Improve the Fault Diagnosis on Gearboxes , 2019, Comput. Intell. Neurosci..

[20]  Yoshua Bengio,et al.  Gradient-based learning applied to document recognition , 1998, Proc. IEEE.

[21]  Moncef Gabbouj,et al.  Real-Time Motor Fault Detection by 1-D Convolutional Neural Networks , 2016, IEEE Transactions on Industrial Electronics.

[22]  K. Gröchenig.  Uncertainty Principles for Time-Frequency Representations , 2003 .

[23]  Philip Chan,et al.  Toward accurate dynamic time warping in linear time and space , 2007, Intell. Data Anal..

[24]  Patrick Schäfer.  The BOSS is concerned with time series classification in the presence of noise , 2014, Data Mining and Knowledge Discovery.

[25]  Diego Cabrera,et al.  Extracting repetitive transients for rotating machinery diagnosis using multiscale clustered grey infogram , 2016 .

[26]  Yonghong Zhang,et al.  Motor Fault Diagnosis Based on Short-time Fourier Transform and Convolutional Neural Network , 2017, Chinese Journal of Mechanical Engineering.

[27]  Robert B. Randall,et al.  The spectral kurtosis: application to the vibratory surveillance and diagnostics of rotating machines , 2006 .

[28]  Jason Lines,et al.  Time-Series Classification with COTE: The Collective of Transformation-Based Ensembles , 2015, IEEE Trans. Knowl. Data Eng..