Improved Boreability Index for Gripper TBMs in Medium- to Strong-Quality Rocks Based on Theoretical Analysis and Field Penetration Tests

Rock Mechanics and Rock Engineering, Год журнала: 2025, Номер unknown

Опубликована: Янв. 31, 2025

Язык: Английский

---

Semi-supervised ensemble model for TBM rock mass classification

Tunnelling and Underground Space Technology, Год журнала: 2025, Номер 162, С. 106632 - 106632

Опубликована: Апрель 14, 2025

Язык: Английский

---

Performance of Fine-Tuning Techniques for Multilabel Classification of Surface Defects in Reinforced Concrete Bridges

Applied Sciences, Год журнала: 2025, Номер 15(9), С. 4725 - 4725

Опубликована: Апрель 24, 2025

Machine learning models often face challenges in bridge inspections, especially handling complex surface features and overlapping defects that make accurate classification difficult. These are common for image-based monitoring, which has become increasingly popular inspecting assessing the structural condition of reinforced concrete bridges with automated possibilities. Despite advances defect detection using convolutional neural networks (CNNs), although such as defects, textures, data imbalance remain difficult, full fine-tuning deep helps them better adapt to these conditions by updating all layers domain-specific learning. The aim this study is demonstrate how effective several architectures damage allows robust performance best utilization value methods. Six CNN architectures, ResNet-18, ResNet-50, ResNet-101, ResNeXt-50, ResNeXt-101 EfficientNet-B3, were fine-tuned CODEBRIM dataset. Their was evaluated Precision, Recall, F1 Score, Balanced Accuracy AUC-ROC metrics ensure a evaluation framework. This indicates EfficientNet-B3 outperformed other achieved highest accuracy error categories. showed best-balanced Precision (0.935) perfect Recall (1.000) background classification, indicating its ability distinguish defect-free areas from damage. results highlight potential improve inspection systems thus increase efficiency real-world applications, well provide guidance selection methods based on whether or overall consistency more important specific application.

Язык: Английский

---

Advanced identification method for adverse geological conditions in TBM tunnels based on stacking ensemble algorithm and Bayesian theory

Tunnelling and Underground Space Technology, Год журнала: 2025, Номер 163, С. 106741 - 106741

Опубликована: Май 13, 2025

Язык: Английский

---