Article
Engineering, Industrial
Naipeng Li, Nagi Gebraeel, Yaguo Lei, Xiaolei Fang, Xiao Cai, Tao Yan
Summary: This paper proposes a RUL prediction method based on a multi-sensor data fusion model to improve prediction performance by prioritizing sensor selection and fusing multi-sensor data. The effectiveness of the method is demonstrated using simulation study and aircraft engine degradation data from NASA repository.
RELIABILITY ENGINEERING & SYSTEM SAFETY
(2021)
Article
Automation & Control Systems
Naipeng Li, Yaguo Lei, Nagi Gebraeel, Zhijian Wang, Xiao Cai, Pengcheng Xu, Biao Wang
Summary: This article proposes a multi-sensor data-driven RUL prediction method for semi-observable systems, leveraging degradation information from online multi-sensor signals as well as offline state observations. The method is developed based on a generalizable state-space model combined with a particle filtering framework, and an algorithm named prioritized sensor group selection is introduced to select the optimal sensor group for RUL prediction, demonstrating its effectiveness in an experiment of cutting tool wear.
IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS
(2021)
Article
Engineering, Mechanical
He Yu, Hongru Li
Summary: This paper proposes a novel method for predicting the remaining useful life (RUL) of hydraulic pumps. The method integrates multiple sources of data and applies monotonicity constraints to improve the prediction performance. Experimental results demonstrate the effectiveness of the proposed approach.
MECHANICAL SYSTEMS AND SIGNAL PROCESSING
(2022)
Article
Automation & Control Systems
Shixiang Lu, Zhiwei Gao, Qifa Xu, Cuixia Jiang, Tianming Xie, Aihua Zhang
Summary: This research proposes an interactive attention-based deep spatio-temporal network to effectively fuse vibration waveforms and time-varying operating signals for predicting remaining useful life of machinery. Experimental results demonstrate that the proposed model outperforms other methods in accurately predicting remaining useful life.
IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS
(2023)
Article
Engineering, Mechanical
Pengfei Liang, Ying Li, Bin Wang, Xiaoming Yuan, Lijie Zhang
Summary: Accurate monitoring of mechanical device conditions requires a large number of sensors working together. Existing deep learning models often focus too much on temporal correlations and ignore spatial correlations of multiple sensors. Therefore, a new end-to-end framework named GAT-DAT is proposed to address these deficiencies, by incorporating a deep adaptative transformer and a graph attention network.
INTERNATIONAL JOURNAL OF FATIGUE
(2023)
Article
Engineering, Industrial
Tianfu Li, Zhibin Zhao, Chuang Sun, Ruqiang Yan, Xuefeng Chen
Summary: This study proposes a sensor network model HAGCN for RUL prediction, which models spatial and temporal dependencies of sensors simultaneously using hierarchical graph representation layer and bi-directional long short-term memory network. Experimental results show the superiority of this method over existing approaches.
RELIABILITY ENGINEERING & SYSTEM SAFETY
(2021)
Article
Computer Science, Hardware & Architecture
Jinhua Mi, Lulu Liu, Yonghao Zhuang, Libing Bai, Yan-Feng Li
Summary: In this article, a synthetic feature processing method is proposed for the remaining useful life prediction of rolling bearings. The method addresses the shortcomings of feature selection, feature fusion, and health state segment, and achieves more accurate and robust prediction results.
IEEE TRANSACTIONS ON RELIABILITY
(2023)
Article
Computer Science, Artificial Intelligence
Xuelin Liu, Baoping Cai, Xiaobing Yuan, Xiaoyan Shao, Yiliu Liu, Javed Akbar Khan, Hongyan Fan, Yonghong Liu, Zengkai Liu, Guijie Liu
Summary: In this study, a hybrid multi-stage methodology for remaining useful life (RUL) prediction of control systems is proposed. The variant of unscented Kalman filter (UKF) and dynamic Bayesian networks (DBNs) are used for uncertainty analysis, and the real degradation process of control systems is simulated by optimizing the degradation process, leading to improved accuracy and robustness of RUL prediction.
EXPERT SYSTEMS WITH APPLICATIONS
(2023)
Article
Green & Sustainable Science & Technology
Chunli Liu, Qiang Li, Kai Wang
Summary: This paper reviews the importance of predicting the remaining useful life and estimating the state-of-charge of supercapacitors for their safe operation, emphasizing the necessity of accurate monitoring and replacement as well as precise state estimation. Various methods are summarized and compared, with a focus on data-based research and the application of artificial neural networks.
RENEWABLE & SUSTAINABLE ENERGY REVIEWS
(2021)
Article
Engineering, Mechanical
Yudong Cao, Minping Jia, Yifei Ding, Xiaoli Zhao, Peng Ding, Liudong Gu
Summary: In the industrial big data context, deep neural networks have been widely used for fault classification and remaining useful life (RUL) prediction of mechanical equipment. However, traditional deep learning models neglect the importance of time-frequency domain analysis for rotating machinery and are limited by single-channel information. To solve these problems, a complex domain extension network with multi-channel information fusion is proposed for RUL prediction. The effectiveness of the proposed framework is demonstrated through case studies using run-to-failure datasets, where it outperforms current state-of-the-art methods in terms of prediction accuracy, interpretability, and generalization.
MECHANICAL SYSTEMS AND SIGNAL PROCESSING
(2023)
Article
Mathematics
Yingzhi Zhang, Guiming Guo, Fang Yang, Yubin Zheng, Fenli Zhai
Summary: A tool remaining useful life prediction method is proposed based on a non-homogeneous Poisson process and Weibull proportional hazard model, considering the grinding repair of machine tools during operation. The method builds an intrinsic failure rate model using tool failure data and establishes a WPHM by collecting vibration information during operation. By incorporating tool grinding repair, the NHPP-WPHM under different repair times is established to describe the tool's comprehensive failure rate. The effectiveness of the model is verified by comparing it with actual remaining useful life and another WPHM-based prediction model.
Article
Engineering, Electrical & Electronic
Kangkai Wu, Jingjing Li, Lichao Meng, Fengling Li, Heng Tao Shen
Summary: Unsupervised domain adaptation (UDA) methods are valuable in cross-domain remaining useful life (RUL) prediction, but many industries value data privacy protection. To address this, this paper proposes a source-free domain adaptation method and utilizes an adversarial architecture to achieve RUL prediction.
IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT
(2023)
Article
Engineering, Multidisciplinary
Jiaxin Yang, Shengjin Tang, Pengya Fang, Fengfei Wang, Xiaoyan Sun, Xiaosheng Si
Summary: In this paper, a RUL prediction method based on the implicit linear Wiener degradation process is proposed to reasonably fuse failure time data or multi-source information, solving the problem of imperfect or scarce prior degradation information in practical engineering applications. The method obtains the relationship between parameter estimation and degradation data through theoretical derivation, estimates fixed parameters using field and historical degradation data, fuses failure time data using the EM algorithm, and updates the drift parameter using the Kalman filtering algorithm.
PROCEEDINGS OF THE INSTITUTION OF MECHANICAL ENGINEERS PART O-JOURNAL OF RISK AND RELIABILITY
(2022)
Article
Engineering, Electrical & Electronic
Yuxuan Zhang, Yuanxiang Li, Yilin Wang, Yongshen Yang, Xian Wei
Summary: Accurate remaining useful life (RUL) prediction is crucial for maintaining the safety and reliability of industrial systems. Deep learning based-methods have shown potential in improving prediction accuracy by learning and fusing degradation features from signals. However, these methods often neglect the characteristics of different sensors in the spatial domain. In this paper, we propose an adaptive spatio-temporal graph neural network (ASTGNN) framework that addresses this issue by considering both spatial structures learning and spatio-temporal information fusion.
IEEE SENSORS JOURNAL
(2022)
Article
Automation & Control Systems
Zian Chen, Xiaohang Jin, Ziqian Kong, Feng Wang, Zhengguo Xu
Summary: This article proposes a novel Transformer-based network to address the unsatisfactory generalization ability and low interpretability of data-driven methods in remaining useful life prediction. By integrating global and local information and utilizing cross-attention, the model considers diverse degradation modes and achieves stronger generalization. The designed cross-attention discrepancy aligns similar degradation states more properly, providing inherent interpretability.
ENGINEERING APPLICATIONS OF ARTIFICIAL INTELLIGENCE
(2023)
