A Novel Transformer-based Few-Shot Learning Method for Intelligent Fault Diagnosis with Noisy Labels under Varying Working Conditions

Reliability Engineering & System Safety, Journal Year: 2024, Volume and Issue: 251, P. 110400 - 110400

Published: July 31, 2024

Language: Английский

---

Multi-source domain adaptation network for partial discharge severity assessment in gas-insulated switchgear

Measurement Science and Technology, Journal Year: 2024, Volume and Issue: 35(12), P. 125105 - 125105

Published: Sept. 9, 2024

Abstract Gas-insulated switchgear (GIS) partial discharge (PD) severity assessment is critical for ensuring the reliable operation of GIS systems. However, existing methods often overlook long-term dependencies historical data and fail to adequately address challenges related limited on-site samples variations in sample distribution. To overcome these challenges, we propose a novel multi-source domain adaptation network (MSDAN) specifically designed PD assessment, which first model developed considering distribution differences different defect types. Our approach begins with development feature extractor that captures both discernible features dependencies. We then introduce strategy mitigate disparities across from types, effective alignment. Additionally, incorporate an adaptive weighted classification mechanism accurately assess by varying contributions types target task. Experimental results demonstrate MSDAN achieves remarkable accuracy 95.38% outperforming other benchmark models robustness. This highlights potential as robust solution real-world assessment.

Language: Английский

---

An Improved Weighted Cross-Entropy-Based Convolutional Neural Network for Auxiliary Diagnosis of Pneumonia

Electronics, Journal Year: 2024, Volume and Issue: 13(15), P. 2929 - 2929

Published: July 24, 2024

Pneumonia has long been a significant concern in global public health. With the advancement of convolutional neural networks (CNNs), new technological methods have emerged to address this challenge. However, application CNNs pneumonia diagnosis still faces several critical issues. First, datasets used for training models often suffer from insufficient sample sizes and imbalanced class distributions, leading reduced classification performance. Second, although can automatically extract features make decisions complex image data, their interpretability is relatively poor, limiting widespread use clinical some extent. To these issues, novel weighted cross-entropy loss function proposed, which calculates weights via an inverse proportion exponential handle data imbalance more efficiently. Additionally, we employ transfer learning approach that combines pretrained CNN model parameter fine-tuning improve Finally, introduce gradient-weighted activation mapping method enhance model’s by visualizing regions focus. The experimental results indicate our proposed significantly enhances performance tasks. Among four selected models, accuracy rates improved over 90%, visualized were provided.

Language: Английский

---

EICD: Causal Structure Recovery of Bearing Failure Data Containing Latent Confounding

IEEE Sensors Journal, Journal Year: 2024, Volume and Issue: 24(19), P. 30560 - 30574

Published: Aug. 27, 2024

Language: Английский

---