Reliable Fault Diagnosis of Bearings Using an Optimized Stacked Variational Denoising Auto-Encoder

Abstract

Variational auto-encoders (VAE) have recently been successfully applied in the intelligent fault diagnosis of rolling bearings due to its self-learning ability and robustness. However, the hyper-parameters of VAEs depend, to a significant extent, on artificial settings, which is regarded as a common and key problem in existing deep learning models. Additionally, its anti-noise capability may face a decline when VAE is used to analyze bearing vibration data under loud environmental noise. Therefore, in order to improve the anti-noise performance of the VAE model and adaptively select its parameters, this paper proposes an optimized stacked variational denoising autoencoder (OSVDAE) for the reliable fault diagnosis of bearings. Within the proposed method, a robust network, named variational denoising auto-encoder (VDAE), is, first, designed by integrating VAE and a denoising auto-encoder (DAE). Subsequently, a stacked variational denoising auto-encoder (SVDAE) architecture is constructed to extract the robust and discriminative latent fault features via stacking VDAE networks layer on layer, wherein the important parameters of the SVDAE model are automatically determined by employing a novel meta-heuristic intelligent optimizer known as the seagull optimization algorithm (SOA). Finally, the extracted latent features are imported into a softmax classifier to obtain the results of fault recognition in rolling bearings. Experiments are conducted to validate the effectiveness of the proposed method. The results of analysis indicate that the proposed method not only can achieve a high identification accuracy for different bearing health conditions, but also outperforms some representative deep learning methods.

Deep-Learning Method Based on 1D Convolutional Neural Network for Intelligent Fault Diagnosis of Rotating Machines

“…Each sample consists of 3000 data points; where a multi-domain feature calculation is applied, the accuracy is 99.8%. Finally, another method reported in [38] has above 99% accuracy rate using an optimized method of stacked variational denoising auto-encoder. In comparison, the proposed model has The method of bearing failure diagnosis has been developed on the basis of feature learning approach.…”

Deep transfer learning rolling bearing fault diagnosis method based on convolutional neural network feature fusion

“…The use of deep learning enables adaptive feature extraction of data and directly complete end-to-end intelligent fault diagnosis, which greatly reduces the need for feature extraction expertise and the uncertainty caused by human involvement [15,16]. Yan et al [17] employed the seagull optimization algorithm to optimize important parameters of a stacked variational denoising self-encoder and effectively extract important features from one-dimensional bearing vibration data. Liu et al [18] proposed a method for extracting fault features using time-domain signals as training data.…”

Research on fault diagnosis of rolling bearings with variable working conditions based on improved residual neural network

“…In bearing fault diagnosis, many researchers have attempted to use different types of deep learning models. For example, improved stacked recurrent neural networks [13], fault recognition methods combining LSTM and transfer learning [14], using CNN and bidirectional gated units to simultaneously learn time-domain and frequency-domain features in the data [15], using autoencoders [16] to mine signal feature information, and adaptive onedimensional convolutional neural networks [17]. These methods have achieved good results, effectively diagnosing and classifying rolling bearing faults and providing reliable guarantees for industrial production.…”