Sparse Autoencoder-based Multi-head Deep Neural Networks for Machinery Fault Diagnostics with Detection of Novelties
Abstract Supervised fault diagnosis typically assumes that all the types of machinery failures are known. However, in practice unknown types of defect, i.e., novelties, may occur, whose detection is a challenging task. In this paper, a novel fault diagnostic method is developed for both diagnostics...
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
SpringerOpen
2021-06-01
|
| Series: | Chinese Journal of Mechanical Engineering |
| Subjects: | |
| Online Access: | https://doi.org/10.1186/s10033-021-00569-0 |
| _version_ | 1818645086357946368 |
|---|---|
| author | Zhe Yang Dejan Gjorgjevikj Jianyu Long Yanyang Zi Shaohui Zhang Chuan Li |
| author_facet | Zhe Yang Dejan Gjorgjevikj Jianyu Long Yanyang Zi Shaohui Zhang Chuan Li |
| author_sort | Zhe Yang |
| collection | DOAJ |
| description | Abstract Supervised fault diagnosis typically assumes that all the types of machinery failures are known. However, in practice unknown types of defect, i.e., novelties, may occur, whose detection is a challenging task. In this paper, a novel fault diagnostic method is developed for both diagnostics and detection of novelties. To this end, a sparse autoencoder-based multi-head Deep Neural Network (DNN) is presented to jointly learn a shared encoding representation for both unsupervised reconstruction and supervised classification of the monitoring data. The detection of novelties is based on the reconstruction error. Moreover, the computational burden is reduced by directly training the multi-head DNN with rectified linear unit activation function, instead of performing the pre-training and fine-tuning phases required for classical DNNs. The addressed method is applied to a benchmark bearing case study and to experimental data acquired from a delta 3D printer. The results show that its performance is satisfactory both in detection of novelties and fault diagnosis, outperforming other state-of-the-art methods. This research proposes a novel fault diagnostics method which can not only diagnose the known type of defect, but also detect unknown types of defects. |
| first_indexed | 2024-12-17T00:25:09Z |
| format | Article |
| id | doaj.art-b89721cf6296483a8a2661e099c0789e |
| institution | Directory Open Access Journal |
| issn | 1000-9345 2192-8258 |
| language | English |
| last_indexed | 2024-12-17T00:25:09Z |
| publishDate | 2021-06-01 |
| publisher | SpringerOpen |
| record_format | Article |
| series | Chinese Journal of Mechanical Engineering |
| spelling | doaj.art-b89721cf6296483a8a2661e099c0789e2022-12-21T22:10:28ZengSpringerOpenChinese Journal of Mechanical Engineering1000-93452192-82582021-06-0134111210.1186/s10033-021-00569-0Sparse Autoencoder-based Multi-head Deep Neural Networks for Machinery Fault Diagnostics with Detection of NoveltiesZhe Yang0Dejan Gjorgjevikj1Jianyu Long2Yanyang Zi3Shaohui Zhang4Chuan Li5School of Mechanical Engineering, Dongguan University of TechnologyFaculty of Computer Science and Engineering, Ss. Cyril and Methodius UniversitySchool of Mechanical Engineering, Dongguan University of TechnologySchool of Mechanical Engineering, Xi’an Jiaotong UniversitySchool of Mechanical Engineering, Dongguan University of TechnologySchool of Mechanical Engineering, Dongguan University of TechnologyAbstract Supervised fault diagnosis typically assumes that all the types of machinery failures are known. However, in practice unknown types of defect, i.e., novelties, may occur, whose detection is a challenging task. In this paper, a novel fault diagnostic method is developed for both diagnostics and detection of novelties. To this end, a sparse autoencoder-based multi-head Deep Neural Network (DNN) is presented to jointly learn a shared encoding representation for both unsupervised reconstruction and supervised classification of the monitoring data. The detection of novelties is based on the reconstruction error. Moreover, the computational burden is reduced by directly training the multi-head DNN with rectified linear unit activation function, instead of performing the pre-training and fine-tuning phases required for classical DNNs. The addressed method is applied to a benchmark bearing case study and to experimental data acquired from a delta 3D printer. The results show that its performance is satisfactory both in detection of novelties and fault diagnosis, outperforming other state-of-the-art methods. This research proposes a novel fault diagnostics method which can not only diagnose the known type of defect, but also detect unknown types of defects.https://doi.org/10.1186/s10033-021-00569-0Deep learningFault diagnosticsNovelty detectionMulti-head deep neural networkSparse autoencoder |
| spellingShingle | Zhe Yang Dejan Gjorgjevikj Jianyu Long Yanyang Zi Shaohui Zhang Chuan Li Sparse Autoencoder-based Multi-head Deep Neural Networks for Machinery Fault Diagnostics with Detection of Novelties Chinese Journal of Mechanical Engineering Deep learning Fault diagnostics Novelty detection Multi-head deep neural network Sparse autoencoder |
| title | Sparse Autoencoder-based Multi-head Deep Neural Networks for Machinery Fault Diagnostics with Detection of Novelties |
| title_full | Sparse Autoencoder-based Multi-head Deep Neural Networks for Machinery Fault Diagnostics with Detection of Novelties |
| title_fullStr | Sparse Autoencoder-based Multi-head Deep Neural Networks for Machinery Fault Diagnostics with Detection of Novelties |
| title_full_unstemmed | Sparse Autoencoder-based Multi-head Deep Neural Networks for Machinery Fault Diagnostics with Detection of Novelties |
| title_short | Sparse Autoencoder-based Multi-head Deep Neural Networks for Machinery Fault Diagnostics with Detection of Novelties |
| title_sort | sparse autoencoder based multi head deep neural networks for machinery fault diagnostics with detection of novelties |
| topic | Deep learning Fault diagnostics Novelty detection Multi-head deep neural network Sparse autoencoder |
| url | https://doi.org/10.1186/s10033-021-00569-0 |
| work_keys_str_mv | AT zheyang sparseautoencoderbasedmultiheaddeepneuralnetworksformachineryfaultdiagnosticswithdetectionofnovelties AT dejangjorgjevikj sparseautoencoderbasedmultiheaddeepneuralnetworksformachineryfaultdiagnosticswithdetectionofnovelties AT jianyulong sparseautoencoderbasedmultiheaddeepneuralnetworksformachineryfaultdiagnosticswithdetectionofnovelties AT yanyangzi sparseautoencoderbasedmultiheaddeepneuralnetworksformachineryfaultdiagnosticswithdetectionofnovelties AT shaohuizhang sparseautoencoderbasedmultiheaddeepneuralnetworksformachineryfaultdiagnosticswithdetectionofnovelties AT chuanli sparseautoencoderbasedmultiheaddeepneuralnetworksformachineryfaultdiagnosticswithdetectionofnovelties |