Article
Automation & Control Systems
Xiang Li, Yixiao Xu, Naipeng Li, Bin Yang, Yaguo Lei
Summary: This paper proposes a deep learning-based method for predicting the remaining useful life (RUL) to address the sensor malfunction problem. By adopting a global feature extraction scheme and introducing adversarial learning, promising and robust RUL prediction performance can be achieved in testing scenarios with sensor malfunctions. The experimental results suggest that the proposed approach is well suited for real industrial applications.
IEEE-CAA JOURNAL OF AUTOMATICA SINICA
(2023)
Article
Automation & Control Systems
Mohamed Ragab, Zhenghua Chen, Min Wu, Chuan Sheng Foo, Chee Keong Kwoh, Ruqiang Yan, Xiaoli Li
Summary: This article introduces a novel contrastive adversarial domain adaptation (CADA) method for cross-domain RUL prediction, which combines adversarial domain adaptation architecture with contrastive loss to effectively consider target-specific information.
IEEE TRANSACTIONS ON INDUSTRIAL INFORMATICS
(2021)
Article
Engineering, Industrial
Yuntian Ta, Yanfeng Li, Wenan Cai, Qianqian Zhang, Zhijian Wang, Lei Dong, Wenhua Du
Summary: This paper proposes an adaptive staged RUL prediction method based on multi-sensor and multi-feature fusion, which improves the accuracy of RUL prediction methods. Firstly, a multi-sensor and multi-feature fusion method is proposed to initially fuse vibration signals and features. Then, an adaptive staged RUL prediction method is proposed to match different degradation models with different degradation stages of the component. The effectiveness and superiority of the proposed method are verified through experiments.
RELIABILITY ENGINEERING & SYSTEM SAFETY
(2023)
Article
Mathematics
Feiyue Deng, Yan Bi, Yongqiang Liu, Shaopu Yang
Summary: This article introduces a multi-scale dilated convolution network (MsDCN) for RUL prediction, which utilizes a new multi-scale dilation convolution fusion unit (MsDCFU) and depthwise separable convolution (DSC) to address the computational efficiency and overfitting issues of traditional multi-scale CNNs.
Article
Engineering, Electrical & Electronic
Kangkai Wu, Jingjing Li, Lin Zuo, Ke Lu, Heng Tao Shen
Summary: The article proposes a weighted adversarial loss (WAL) for cross-domain RUL prediction, which effectively enhances positive transfer and alleviates negative transfer, achieving state-of-the-art results.
IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT
(2022)
Article
Automation & Control Systems
Meng Ma, Zhu Mao
Summary: Accurate prediction of remaining useful life (RUL) is critical in the field of prognostics and health management (PHM). A novel deep neural network named CLSTM is proposed in this article, which combines convolutional operation with LSTM to improve prediction accuracy and computation efficiency for RUL of rotating machineries. Results show that the proposed CLSTM network outperforms current deep learning algorithms in URL prediction and system prognosis.
IEEE TRANSACTIONS ON INDUSTRIAL INFORMATICS
(2021)
Article
Engineering, Chemical
Wei Hao, Zhixuan Li, Guohao Qin, Kun Ding, Xuwei Lai, Kai Zhang
Summary: This study proposed a novelty RUL prediction model for rolling bearings based on a bi-channel hierarchical vision transformer to reduce the impact of the above problems on prediction accuracy improvement.
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
Engineering, Electrical & Electronic
Tiancheng Wang, Di Guo, Xi-Ming Sun
Summary: Recent research has shown the potential of deep learning algorithms in improving industrial automation for prognostics of remaining useful life (RUL). However, current continual learning methods for RUL prediction are not effective in noisy and diverse problems. This article proposes a contrastive generative replay (GR) method that uses contrastive learning to improve the effectiveness of continual learning and demonstrates its efficacy in continuous RUL prediction for bearings.
IEEE SENSORS JOURNAL
(2023)
Article
Computer Science, Artificial Intelligence
Gang Wang, Hui Li, Feng Zhang, Zhangjun Wu
Summary: This paper proposes a Feature Fusion based Ensemble Method (FFEM) for predicting the Remaining Useful Life (RUL) of machinery. The method utilizes the characteristics of signal analysis features and deep representation features, and combines different types of features using a fusion method. Experimental results on a run-to-failure dataset of bearings demonstrate the effectiveness of the proposed method.
APPLIED SOFT COMPUTING
(2022)
Article
Engineering, Aerospace
Xiaofeng Liu, Liuqi Xiong, Yiming Zhang, Chenshuang Luo, Sujin Bureerat
Summary: This paper proposes a residual life prediction model based on Autoencoder and TCN, which reduces the data dimension and extracts features to predict the remaining useful life of the engine. The experimental results show that the proposed model performs the best in evaluation, and it has important implications for engine health.
Article
Engineering, Industrial
Zhifu Huang, Yang Yang, Yawei Hu, Xiang Ding, Xuanlin Li, Yongbin Liu
Summary: Deep learning methods are increasingly important in RUL prediction for machines due to their powerful nonlinear mapping capabilities. However, these methods often suffer from information leakage and correlation loss between features and data during the mapping process. A novel attention-augmented recalibrated and compensatory network (ATRCN) is proposed, which strengthens the correlation between features and attention weights, and compensates for information leakage through a global compensation-information mechanism. Experimental results show that ATRCN outperforms existing approaches.
RELIABILITY ENGINEERING & SYSTEM SAFETY
(2023)
Article
Automation & Control Systems
Hui Liu, Zhenyu Liu, Weiqiang Jia, Xianke Lin
Summary: This article introduces a novel feature-attention-based end-to-end approach for RUL prediction, which effectively improves the model's predictive performance. By dynamically allocating attention weights in the input data and combining the use of BGRU and convolutional neural networks for extracting long-term dependencies and capturing local features, abstract representations for RUL prediction are learned.
IEEE TRANSACTIONS ON INDUSTRIAL INFORMATICS
(2021)
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
Energy & Fuels
Muhammad Haris, Muhammad Noman Hasan, Shiyin Qin
Summary: This study introduces a novel method to predict the remaining useful life (RUL) of supercapacitors in the early stages of degradation, by combining deep learning algorithm with Bayesian optimization. The proposed method successfully reduces the time and resources required for developing the RUL prediction model while maintaining good accuracy of the model.