Deep learning and its applications to machine health monitoring

Applications of machine learning to machine fault diagnosis: A review and roadmap

With the rapid development of smart manufacturing, data-driven fault diagnosis has attracted increasing attentions. As one of the most popular methods applied in fault diagnosis, deep learning (DL) has achieved remarkable results. However, due to the fact that the volume of labeled samples is small in fault diagnosis, the depths of DL models for fault diagnosis are shallow compared with convolutional neural network in other areas (including ImageNet), which limits their final prediction accuracies. In this research, a new TCNN(ResNet-50) with the depth of 51 convolutional layers is proposed for fault diagnosis. By combining with transfer learning, TCNN(ResNet-50) applies ResNet-50 trained on ImageNet as feature extractor for fault diagnosis. Firstly, a signal-to-image method is developed to convert time-domain fault signals to RGB images format as the input datatype of ResNet-50. Then, a new structure of TCNN(ResNet-50) is proposed. Finally, the proposed TCNN(ResNet-50) has been tested on three datasets, including bearing damage dataset provided by KAT datacenter, motor bearing dataset provided by Case Western Reserve University (CWRU) and self-priming centrifugal pump dataset. It achieved state-of-the-art results. The prediction accuracies of TCNN(ResNet-50) are as high as 98.95% ± 0.0074, 99.99% ± 0 and 99.20% ± 0, which demonstrates that TCNN(ResNet-50) outperforms other DL models and traditional methods.

A transfer convolutional neural network for fault diagnosis based on ResNet-50

Vibration based condition monitoring and fault diagnosis of wind turbine planetary gearbox: A review

In this survey paper, we systematically summarize existing literature on bearing fault diagnostics with deep learning (DL) algorithms. While conventional machine learning (ML) methods, including artificial neural network, principal component analysis, support vector machines, etc., have been successfully applied to the detection and categorization of bearing faults for decades, recent developments in DL algorithms in the last five years have sparked renewed interest in both industry and academia for intelligent machine health monitoring. In this paper, we first provide a brief review of conventional ML methods, before taking a deep dive into the state-of-the-art DL algorithms for bearing fault applications. Specifically, the superiority of DL based methods are analyzed in terms of fault feature extraction and classification performances; many new functionalities enabled by DL techniques are also summarized. In addition, to obtain a more intuitive insight, a comparative study is conducted on the classification accuracy of different algorithms utilizing the open source Case Western Reserve University (CWRU) bearing dataset. Finally, to facilitate the transition on applying various DL algorithms to bearing fault diagnostics, detailed recommendations and suggestions are provided for specific application conditions. Future research directions to further enhance the performance of DL algorithms on health monitoring are also discussed.

Deep Learning Algorithms for Bearing Fault Diagnosticsx—A Comprehensive Review

Matching and reassignment based time-frequency enhancement for rotating machinery fault diagnosis under nonstationary speed operations

Graph deep learning (DL)-based prognostic methods have been successfully applied in bearing remaining useful life (RUL) prediction, as graph represents spatial and temporal dependencies of signals. However, graph data-driven prediction methods using single-sensor data are still insufficiently studied, and the graph construction is not interpretable, where the physical meaning of edges is unclear. To overcome these limitations, a node-level PathGraph-based bearing RUL prediction method is proposed, where a Chebyshev graph convolutional network (ChebCGN) with bidirectional long short-term memory network (BiLSTM) is designed. The node-level PathGraph is constructed to represent the relationships among the time-discrete signals, where edges denote the chronological order and nodes represent the signals. After that, graph feature learning ability of ChebGCN-LSTM is enhanced by inputting different chronological PathGraphs related to bearings’ states. In ChebGCN-LSTM, the BiLSTM captures the temporal information, overcoming the limitation of ChebGCN that ignored global temporal dependencies of signals. The constructed PathGraphs are input to ChebGCN-LSTM simultaneously to realize RUL prediction. Experimental results on case studies verify the effectiveness of the proposed bearing RUL prediction method.

A Node-Level PathGraph-Based Bearing Remaining Useful Life Prediction Method

The utility model relates to a self -adaptation degree of depth confidence network bearing failure diagnostic device based on nesterov momentum method, including the signal acquisition module for acquire the different health condition's of antifriction bearing original information, the failure diagnosis module, with the signal acquisition module is connected, regards the original information as incoming signal, carries out automatic withdrawal to the deep characteristic of original information, unknown status signal input module, with the failure diagnosis module is connected, by antifriction bearing's operating condition and fault type are judged to the failure diagnosis module. The utility model discloses on the basis of rolling bearing fault classification, to the trouble classification accuracy limited problem of present degree of depth confidence network (Deep belief network) to the original information, add nesterov momentum method and independent self -adaptation study rate for DBN trains speed in advance, improves the trouble classification accuracy.

Self -adaptation degree of depth confidence network bearing failure diagnostic device based on nesterov momentum method

The displacement feedback with time delay considered is introduced in order to enhance the vibration isolation performance of a high-static-low-dynamic stiffness (HSLDS) vibration isolator. Such feedback is detailedly analyzed from the viewpoint of equivalent damping. Firstly, the primary resonance of the controlled HSLDS vibration isolator subjected to a harmonic force excitation is obtained based on the multiple scales method and further verified by numerical integration. The stability of the primary resonance is subsequently investigated. Then, the equivalent damping is defined to study the effects of feedback gain and time delay on primary resonance. The condition of jump avoidance is obtained with the purpose of eliminating the adverse effects induced by jumps. Finally, the force transmissibility of the controlled HSLDS vibration isolator is defined to evaluate its isolation performance. It is shown that an appropriate choice of feedback parameters can effectively suppress the force transmissibility in resonant region and reduce the resonance frequency. Furthermore, a wider vibration isolation frequency bandwidth can be achieved compared to the passive HSLDS vibration isolator.

Performance analysis of high-static-low-dynamic stiffness vibration isolator with time-delayed displacement feedback

Mechanical fault diagnosis is essential in ensuring the safety of production and economic development. In the field of fault diagnosis, deep learning has been extensively used due to its excellent feature learning ability. However, it still suffers from several issues; for example, 1) simultaneous requirements of features from multiple aspects, including sparsity and robustness, are hardly met due to the limited feature learning ability of a single model, and 2) most methods deal with preprocessed signals instead of original time domain signals because of the noise interference and deficiency of a single model. To solve these problems, this study proposes a new deep fusion network for fault feature learning, which combines two types of deep learning models, namely, sparse autoencoder and contractive autoencoder, which are respectively applied to enhance features' sparsity and robustness and thereby guarantee the representativeness of extracted features and gain strong anti-interference ability. Consequently, fault diagnosis with original time domain signals can be realized. Bearing and gearbox fault diagnosis experiments are conducted to verify the performance of the presented network. Results show that the diagnosis accuracies for two cases are higher than those of networks based on single contractive autoencoder and sparse autoencoder. These results demonstrate that the proposed fusion network has superior feature learning ability relative to single model networks and can deal with original time domain signals by simultaneously enhancing features' sparsity and robustness.

/pdf/a-new-deep-fusion-network-for-automatic-mechanical-fault-4fgk3616w3.pdf

Xingxing Jiang

Papers

Matching and reassignment based time-frequency enhancement for rotating machinery fault diagnosis under nonstationary speed operations

A Node-Level PathGraph-Based Bearing Remaining Useful Life Prediction Method

Self -adaptation degree of depth confidence network bearing failure diagnostic device based on nesterov momentum method

Performance analysis of high-static-low-dynamic stiffness vibration isolator with time-delayed displacement feedback

A New Deep Fusion Network for Automatic Mechanical Fault Feature Learning