Non-Mutually Exclusive Deep Neural Network Classifier for Combined Modes of Bearing Fault Diagnosis
Open Access
- 7 April 2018
- Vol. 18 (4), 1129
- https://doi.org/10.3390/s18041129
Abstract
The simultaneous occurrence of various types of defects in bearings makes their diagnosis more challenging owing to the resultant complexity of the constituent parts of the acoustic emission (AE) signals. To address this issue, a new approach is proposed in this paper for the detection of multiple combined faults in bearings. The proposed methodology uses a deep neural network (DNN) architecture to effectively diagnose the combined defects. The DNN structure is based on the stacked denoising autoencoder non-mutually exclusive classifier (NMEC) method for combined modes. The NMEC-DNN is trained using data for a single fault and it classifies both single faults and multiple combined faults. The results of experiments conducted on AE data collected through an experimental test-bed demonstrate that the DNN achieves good classification performance with a maximum accuracy of 95%. The proposed method is compared with a multi-class classifier based on support vector machines (SVMs). The NMEC-DNN yields better diagnostic performance in comparison to the multi-class classifier based on SVM. The NMEC-DNN reduces the number of necessary data collections and improves the bearing fault diagnosis performance.Keywords
This publication has 23 references indexed in Scilit:
- A Sequential k-Nearest Neighbor Classification Approach for Data-Driven Fault Diagnosis Using Distance- and Density-Based Affinity MeasuresLecture Notes in Computer Science, 2016
- Fault detection in reciprocating compressor valves under varying load conditionsMechanical Systems and Signal Processing, 2016
- Multifault Diagnosis of Rolling Element Bearings Using a Wavelet Kurtogram and Vector Median-Based Feature AnalysisShock and Vibration, 2015
- Time-Varying and Multiresolution Envelope Analysis and Discriminative Feature Analysis for Bearing Fault DiagnosisIEEE Transactions on Industrial Electronics, 2015
- Application of empirical mode decomposition and artificial neural network for automatic bearing fault diagnosis based on vibration signalsApplied Acoustics, 2015
- High-Performance and Energy-Efficient Fault Diagnosis Using Effective Envelope Analysis and Denoising on a General-Purpose Graphics Processing UnitIEEE Transactions on Power Electronics, 2014
- A segmental hidden semi-Markov model (HSMM)-based diagnostics and prognostics framework and methodologyMechanical Systems and Signal Processing, 2007
- Best basis-based intelligent machine fault diagnosisMechanical Systems and Signal Processing, 2005
- An Amplitude Modulation Detector for Fault Diagnosis in Rolling Element BearingsIEEE Transactions on Industrial Electronics, 2004
- Basic vibration signal processing for bearing fault detectionIEEE Transactions on Education, 2003