Rolling bearing performance assessment with degradation twin modeling considering interdependent fault evolution

Mechanical Systems and Signal Processing, Год журнала: 2024, Номер 224, С. 112194 - 112194

Опубликована: Дек. 6, 2024

Язык: Английский

---

An aircraft engine remaining useful life prediction method based on predictive vector angle minimization and feature fusion gate improved transformer model

Journal of Manufacturing Systems, Год журнала: 2024, Номер 76, С. 567 - 584

Опубликована: Сен. 1, 2024

Язык: Английский

---

A multi-scale cross-dimension interaction approach with adaptive dilated TCN for RUL prediction

Scientific Reports, Год журнала: 2025, Номер 15(1)

Опубликована: Июнь 2, 2025

In the domain of Prognostics and Health Management (PHM) technologies, focus Remaining Useful Life (RUL) prediction is on forecasting time to failure by uncovering complex correlations between equipment degradation features RUL labels, thereby enabling effective support for predictive maintenance strategies. However, extant research primarily emphasizes single-scale single-dimensional feature extraction, which fails adequately capture both long- short-term dependencies as well interrelationships among sensor dimensions. This limitation has a detrimental effect accuracy robustness prediction. To address aforementioned issues, this paper proposes an Adaptive Dilated Temporal Convolutional Network (AD-TCN) approach, incorporating Multi-Scale Cross-Dimension Interaction Module (MSCDIM) enhance extraction interaction. First, dynamic adaptive dilation factor incorporated into TCN, model adjust its receptive field dynamically, facilitates across different scales, allowing more comprehensive representation patterns. Second, MSCDIM module been designed perform multi-scale interaction, fusion temporal dimensions, dynamically adjusting weights in order suppress redundant information critical features. Finally, contrastive ablation experiments are conducted widely used C-MAPSS dataset N-CMAPSS dataset. And experimental results demonstrate that proposed method achieves high performance, with exhibiting strong adaptability significant potential broad applications.

Язык: Английский

---

A hybrid network TEdgeNeXt for data-limited and resource-constrained fault diagnosis

Journal of Vibration and Control, Год журнала: 2024, Номер unknown

Опубликована: Июль 23, 2024

In the field of intelligent machinery fault diagnosis, overcoming challenges arising from scarce labeled data and demand for deployment on resource-constrained edge devices is imperative. To address these hurdles, this work aims to improve ability deep learning models learn strong feature representations limited data, while also reducing model complexity. We presenting a novel network named TEdgeNeXt, approach begins with new signal-to-image conversion method, which proved be able acquire less training quantity. Structurally, Convolutional (Conv.) Encoder initially employed depth-wise separable convolution control size rather than traditional convolution, Split Depth-wise Transpose Attention (SDTA) encoder consequently utilized by leveraging multidimensional processing Multi-head Self-Attention across channel dimensions instead spatial channel. By doing so, it effectively handles such as high multiply-additions (MAdds) increased latency through Flops params. On other hand, fine-tune-based transfer technique extended in our improving capacity generalizing. Ultimately, indicates noticeable improvements Top-1 Accuracy (T1A), Mean Precision (MP), Recall (MR), F1 score (MF1) three distinct datasets.

Язык: Английский

---

PSTFormer: A novel parallel spatial-temporal transformer for remaining useful life prediction of aeroengine

Expert Systems with Applications, Год журнала: 2024, Номер unknown, С. 125995 - 125995

Опубликована: Дек. 1, 2024

Язык: Английский

---

A teacher–student strategy specific to transformer for machine fault diagnosis

Structural Health Monitoring, Год журнала: 2024, Номер unknown

Опубликована: Дек. 25, 2024

To address the challenges of limited labeled data and computational resources in intelligent machine fault diagnosis, we propose a teacher–student strategy based on transformers with token distillation. This approach introduces learnable embedding attention mechanism, enabling student network to learn diagnostic features from larger teacher network. is especially beneficial when model applied new operating conditions via joint classifier. Building Vision Transformer architecture, known for its success large-scale image datasets, our method starts by converting signals into samples. Soft distillation using mechanisms then facilitates Transformer’s training data. In addition, pretraining comprehensive mechanical dataset diverse types improves model’s performance specific target allowing it generalize unseen faults. demonstrates strong Top-1 accuracy, mean precision, recall, F1 score across datasets involving bearings, gears, rotors, improving accuracy even are scarce.

Язык: Английский

---

Single Well Production Prediction Model of Gas Reservoir Based on CNN-BILSTM-AM

Energies, Год журнала: 2024, Номер 17(22), С. 5674 - 5674

Опубликована: Ноя. 13, 2024

In the prediction of single-well production in gas reservoirs, traditional empirical formula reservoirs generally shows poor accuracy. process machine learning training and prediction, problems small data volume dirty are often encountered. order to overcome above problems, a model based on CNN-BILSTM-AM is proposed. The built by long-term short-term memory neural networks, convolutional networks attention modules. input includes previous period its influencing factors. At same time, fitting error value reservoir introduced predict future data. loss function used evaluate deviation between predicted real data, Bayesian hyperparameter optimization algorithm optimize structure comprehensively improve generalization ability model. Three single wells Daniudi D28 well area were selected as database, was production. results show that compared with network (CNN) model, long (LSTM) bidirectional (BILSTM) test set three experimental reduced 6.2425%, 4.9522% 3.0750% average. It basis coupling meets high-precision requirements for which great significance guide efficient development oil fields ensure safety China’s energy strategy.

Язык: Английский

---

A feature separation transfer network with contrastive metric for remaining useful life prediction under different working conditions

Reliability Engineering & System Safety, Год журнала: 2024, Номер unknown, С. 110790 - 110790

Опубликована: Дек. 1, 2024

Язык: Английский

---

A novel remaining useful life prediction method under multiple operating conditions based on attention mechanism and deep learning

Advanced Engineering Informatics, Год журнала: 2024, Номер 64, С. 103083 - 103083

Опубликована: Дек. 27, 2024

Язык: Английский

---