Multi-Fault Bearing Classification Using Sensors and ConvNet-Based Transfer Learning Approach

Udmale, Sandeep S.; Singh, Sanjay Kumar; Singh, Rishav; Sangaiah, Arun Kumar

doi:10.1109/jsen.2019.2947026

Cited by 45 publications

(23 citation statements)

References 32 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Transfer learning refers to a learning process that uses the similarity between data, tasks, or models to apply the models learned in the old domain (source domain) to the new domain (target domain). According to whether the samples in the target domain are labeled or not, transfer learning can be divided into the following categories: 1) supervised transfer learning, 2) semi-supervised transfer learning, and 3) unsupervised transfer learning [32]. We focus on addressing the fault diagnosis problem wherein the target domain has no labeled samples and different data distribution from the source domain because of the variable load of the rotating machinery.…”

Section: Unsupervised Transfer Learningmentioning

confidence: 99%

Multi-scale deep intra-class transfer learning for bearing fault diagnosis

Wang

Shen

Xia

et al. 2020

Reliability Engineering & System Safety

224

View full text Add to dashboard Cite

The tremendous success of deep learning in machine fault diagnosis is dependent on the hypothesis that training and test datasets are subordinated to the same distribution. This subordination is difficult to meet in practical scenarios of industrial applications. On the one hand, the working conditions of rotating machinery can change easily. On the other hand, vibration data and labels are difficult to obtain to train a specific model for each working condition. In this study, we solve these problems by constructing a novel deep transfer learning model called multi-scale deep intra-class adaptation network, which first uses the modified ResNet-50 to extract low-level features and then constructs a multiple scale feature learner to analyze these low-level features at multiple scales and obtain high-level features as input for the classifier. Pseudo labels are then computed to shorten the conditional distribution distance of vibration data collected under different working loads for intra-class adaptation. The proposed method is validated using two datasets to recognize the bearing normal state, the inner race, the ball and outer race faults, and their fault degrees under four different working loads. The high-precision diagnosis results of 24 transfer learning experiments reveal the reliability and generalizability of the constructed model.

show abstract

Section: Unsupervised Transfer Learningmentioning

confidence: 99%

Multi-scale deep intra-class transfer learning for bearing fault diagnosis

Wang

Shen

Xia

et al. 2020

Reliability Engineering & System Safety

224

View full text Add to dashboard Cite

show abstract

“…With an appropriate network architecture with gradient based backpropagation, CNNs synthesize a complex decision surface and are capable of classifying high-dimensional patterns [9]. Researchers are exploring ensemble transfer CNNs [10], [11], including implementations based on stochastic pooling and Leaky Rectified Linear Unit (LReLU) on multichannel signals for fault diagnosis [12]. However, such complex models are incompatible with edge devices due to memory and processing speed constraints [13] and hence most diagnostic solutions utilize cloud-based post-processing.…”

Section: Introductionmentioning

confidence: 99%

Light-Weight CNN Enabled Edge-Based Framework for Machine Health Diagnosis

Mukherjee

Tallur

2021

IEEE Access

View full text Add to dashboard Cite

Condition Based Monitoring (CBM) leverages sensor measurements for measuring state of health of an asset and autonomous diagnosis of faults to trigger remedial actions. Countless deep learning architectures are available for cloud-based feature engineering, feature extraction and classification of data for CBM models. However, complex models pose memory and processing speed constraints for edge implementation, while cloud-based computing poses high latency, high cost of data transmission and storage and privacy threats. This calls for a data driven machine health diagnosis system that is effective yet edge-compatible and secure. In this work, we present a model that is light-weight and non-redundant by optimizing the model size and complexity for edge implementation onto resource constrained, low-cost hardware. The model is based on a light-weight, edge implementable Convolutional Neural Network (CNN) algorithm that utilizes vibration sensor measurements for fault event estimation of machines. The model was trained and tested on two publicly available and widely studied vibration datasets for rolling element bearing faults. This CNN based classification system using spectrograms is able to achieve near perfect accuracy for both binary as well as multi-class fault classification. Lastly, the trained model was implemented and tested on a Raspberry Pi single-board computer, thus realizing an edge-compatible implementation which is cost-saving, secure and reliable.INDEX TERMS Condition based monitoring, convolutional neural network, machine health monitoring, edge computing, vibration analysis

show abstract

“…Guo et al [ 19 ] proposed deep transfer learning-based methods using maximum mean discrepancy and adversarial training techniques together to regularize the discrepancy between different domains. Sandeep et al [ 20 ] presented a ConvNet-based transfer learning method for bearing fault diagnosis with varying speeds. Hasan et al [ 21 ] proposed a transfer learning fault diagnosis framework using 2D acoustic spectral imaging-based pattern formation method.…”

Section: Introductionmentioning

confidence: 99%

WDA: An Improved Wasserstein Distance-Based Transfer Learning Fault Diagnosis Method

Zhu

Wang

Peng

et al. 2021

Sensors

View full text Add to dashboard Cite

With the growth of computing power, deep learning methods have recently been widely used in machine fault diagnosis. In order to realize highly efficient diagnosis accuracy, people need to know the detailed health condition of collected signals from equipment. However, in the actual situation, it is costly and time-consuming to close down machines and inspect components. This seriously impedes the practical application of data-driven diagnosis. In comparison, the full-labeled machine signals from test rigs or online datasets can be achieved easily, which is helpful for the diagnosis of real equipment. Thus, we introduced an improved Wasserstein distance-based transfer learning method (WDA), which learns transferable features between labeled and unlabeled signals from different forms of equipment. In WDA, Wasserstein distance with cosine similarity is applied to narrow the gap between signals collected from different machines. Meanwhile, we use the Kuhn–Munkres algorithm to calculate the Wasserstein distance. In order to further verify the proposed method, we developed a set of case studies, including two different mechanical parts, five transfer scenarios, and eight transfer learning fault diagnosis experiments. WDA reached an average accuracy of 93.72% in bearing fault diagnosis and 84.84% in ball screw fault diagnosis, which greatly surpasses state-of-the-art transfer learning fault diagnosis methods. In addition, comprehensive analysis and feature visualization are also presented.

show abstract

Multi-Fault Bearing Classification Using Sensors and ConvNet-Based Transfer Learning Approach

Cited by 45 publications

References 32 publications

Multi-scale deep intra-class transfer learning for bearing fault diagnosis

Multi-scale deep intra-class transfer learning for bearing fault diagnosis

Light-Weight CNN Enabled Edge-Based Framework for Machine Health Diagnosis

WDA: An Improved Wasserstein Distance-Based Transfer Learning Fault Diagnosis Method

Contact Info

Product

Resources

About