| JIANG Zhinong (江志农)* **,CHEN Yuyang* **,ZHANG Jinjie* **,LI Zhaoyang* **,MAO Zhiwei* **,ZHI Haifeng***,LIU Fengchun***.[J].高技术通讯(英文),2023,29(1):12~21 |
|
| Research on mechanical wear life feature fusion prediction method based on temporal pattern attention mechanism |
| |
| DOI：10. 3772/ j. issn. 1006-6748. 2023. 01. 002 |
| 中文关键词: |
| 英文关键词: prediction of gear remaining useful life, information fusion, numerical simulation, neural network, oil monitoring |
| 基金项目: |
| Author Name | Affiliation | | JIANG Zhinong (江志农)* ** | (*Beijing Key Laboratory of Health Monitoring and Self Healing of High end Mechanical Equipment,
Beijing University of Chemical Technology, Beijing 100029, P.R.China)
(**Key Laboratory of Engine Health Monitoring control and Networking, Ministry of Education,
Beijing University of Chemical Technology, Beijing 100029, P.R.China)
(***China North Engine Research Institute (Tianjin), Tianjin 300400, P.R.China) | | CHEN Yuyang* ** | (*Beijing Key Laboratory of Health Monitoring and Self Healing of High end Mechanical Equipment,
Beijing University of Chemical Technology, Beijing 100029, P.R.China)
(**Key Laboratory of Engine Health Monitoring control and Networking, Ministry of Education,
Beijing University of Chemical Technology, Beijing 100029, P.R.China)
(***China North Engine Research Institute (Tianjin), Tianjin 300400, P.R.China) | | ZHANG Jinjie* ** | (*Beijing Key Laboratory of Health Monitoring and Self Healing of High end Mechanical Equipment,
Beijing University of Chemical Technology, Beijing 100029, P.R.China)
(**Key Laboratory of Engine Health Monitoring control and Networking, Ministry of Education,
Beijing University of Chemical Technology, Beijing 100029, P.R.China)
(***China North Engine Research Institute (Tianjin), Tianjin 300400, P.R.China) | | LI Zhaoyang* ** | (*Beijing Key Laboratory of Health Monitoring and Self Healing of High end Mechanical Equipment,
Beijing University of Chemical Technology, Beijing 100029, P.R.China)
(**Key Laboratory of Engine Health Monitoring control and Networking, Ministry of Education,
Beijing University of Chemical Technology, Beijing 100029, P.R.China)
(***China North Engine Research Institute (Tianjin), Tianjin 300400, P.R.China) | | MAO Zhiwei* ** | (*Beijing Key Laboratory of Health Monitoring and Self Healing of High end Mechanical Equipment,
Beijing University of Chemical Technology, Beijing 100029, P.R.China)
(**Key Laboratory of Engine Health Monitoring control and Networking, Ministry of Education,
Beijing University of Chemical Technology, Beijing 100029, P.R.China)
(***China North Engine Research Institute (Tianjin), Tianjin 300400, P.R.China) | | ZHI Haifeng*** | (*Beijing Key Laboratory of Health Monitoring and Self Healing of High end Mechanical Equipment,
Beijing University of Chemical Technology, Beijing 100029, P.R.China)
(**Key Laboratory of Engine Health Monitoring control and Networking, Ministry of Education,
Beijing University of Chemical Technology, Beijing 100029, P.R.China)
(***China North Engine Research Institute (Tianjin), Tianjin 300400, P.R.China) | | LIU Fengchun*** | (*Beijing Key Laboratory of Health Monitoring and Self Healing of High end Mechanical Equipment,
Beijing University of Chemical Technology, Beijing 100029, P.R.China)
(**Key Laboratory of Engine Health Monitoring control and Networking, Ministry of Education,
Beijing University of Chemical Technology, Beijing 100029, P.R.China)
(***China North Engine Research Institute (Tianjin), Tianjin 300400, P.R.China) |
|
| Hits: 604 |
| Download times: 455 |
| 中文摘要: |
| |
| 英文摘要: |
| In order to solve the problem of low prediction accuracy when only vibration or oil signal is used to predict the remaining life of gear wear, a gear wear life feature fusion prediction method based on temporal pattern attention mechanism is proposed. Firstly, deep residual shrinkage network (DRSN) is used to extract the features of the original vibration time series signals with low signal-to-noise ratio, and the vibration features associated with gear wear evolution are obtained. Secondly, the extracted vibration features and the oil monitoring data that can intuitively reflect the wear process information are jointly input into the bi-directional long short-term memory neural network based on temporal pattern attention mechanism (TPA-BiLSTM), the complex nonlinear relationship between vibration features, oil features and gear wear process evolution is further explored to improve the prediction accuracy. The gear life cycle dynamic response and wear process signals are obtained based on the gear numerical simulation model, and the feasibility of the proposed method is verified. Finally, the proposed method is applied to the residual life prediction of gear on a test bench, and the comparison between different methods proved the validity of the proposed method. |
|
View Full Text
View/Add Comment Download reader |
| Close |
|
|
|
