Back to Search
Start Over
Real-time classification model for tunnel surrounding rocks based on high-resolution neural network and structure–optimizer hyperparameter optimization.
Real-time classification model for tunnel surrounding rocks based on high-resolution neural network and structure–optimizer hyperparameter optimization.
- Source :
-
Computers & Geotechnics . Apr2024, Vol. 168, pN.PAG-N.PAG. 1p. - Publication Year :
- 2024
-
Abstract
- An accurate and real-time identification of the quality grades of tunnel surrounding rocks is essential for efficient tunneling and geological hazard prevention when using tunnel boring machines (TBMs). In this study, an intelligent surrounding rock classification (SRC) model based on TBM operational parameters was constructed using the simultaneous optimization (SO) of the network structure and optimizer hyperparameters. First, a high-quality database for SRC was constructed after abnormal data elimination, data correlation analysis, and data discretization. Six input parameters were selected: advance rate, total thrust, cutterhead torque, cutterhead rotational speed, and pitch and roll angles of the gripper shoes. Second, the SO method was proposed using the Bayesian optimization (BO) algorithm to search for the optimal structure and optimizer hyperparameters of the high-resolution neural network (HRNet), and an HRNet–BO model was constructed for SRC. The evaluation results of the accuracy, precision, recall, micro-F1 score, macro-F1 score, and receiver operating characteristic (ROC) curve showed that the HRNet–BO model had greater robustness in learning and generalizing than the VGG–BO and ResNet–BO models. The proposed SO method and HRNet–BO model can effectively support TBM tunneling and real-time SRC. [ABSTRACT FROM AUTHOR]
Details
- Language :
- English
- ISSN :
- 0266352X
- Volume :
- 168
- Database :
- Academic Search Index
- Journal :
- Computers & Geotechnics
- Publication Type :
- Academic Journal
- Accession number :
- 175872531
- Full Text :
- https://doi.org/10.1016/j.compgeo.2024.106155