The ability to predict tunnel deformation holds great significance for ensuring the reliability, safety, and sustainability of tunnel struc-tures. However, existing deformation prediction models often simplify or over...The ability to predict tunnel deformation holds great significance for ensuring the reliability, safety, and sustainability of tunnel struc-tures. However, existing deformation prediction models often simplify or overlook the impact of spatial characteristics on deformationby treating it as a time series prediction issue. This study utilizes monitoring data from the Grand Canyon Tunnel and introduces aneffective data-driven method for predicting tunnel deformation based on the spatio-temporal characteristics of the historical deformationof adjacent sections. The proposed model, a combination of graph attention network (GAT) and bidirectional long and short-term mem-ory network (Bi-LSTM), is equipped with robust spatio-temporal predictive capabilities. Additionally, the study explores other possiblespatial connections and the scalability of the model. The results indicate that the proposed model outperforms other deep learning mod-els, achieving favorable root mean square error (RMSE), mean absolute error (MAE), and coefficient of determination (R^(2)) values of0.34 mm, 0.23 mm, and 0.94, respectively. The graph structure based on intuitive spatial connections proves more suitable for meetingthe challenges of predicting deformation. Integrating GAT-LSTM with transfer learning technology, remains stable performance whenextended to other tunnels with limited data.展开更多
基金supported by the Sichuan Pro-vince Natural Science Foundation Innovative Research Group Project(Grant No.2024NSFTD0013).
文摘The ability to predict tunnel deformation holds great significance for ensuring the reliability, safety, and sustainability of tunnel struc-tures. However, existing deformation prediction models often simplify or overlook the impact of spatial characteristics on deformationby treating it as a time series prediction issue. This study utilizes monitoring data from the Grand Canyon Tunnel and introduces aneffective data-driven method for predicting tunnel deformation based on the spatio-temporal characteristics of the historical deformationof adjacent sections. The proposed model, a combination of graph attention network (GAT) and bidirectional long and short-term mem-ory network (Bi-LSTM), is equipped with robust spatio-temporal predictive capabilities. Additionally, the study explores other possiblespatial connections and the scalability of the model. The results indicate that the proposed model outperforms other deep learning mod-els, achieving favorable root mean square error (RMSE), mean absolute error (MAE), and coefficient of determination (R^(2)) values of0.34 mm, 0.23 mm, and 0.94, respectively. The graph structure based on intuitive spatial connections proves more suitable for meetingthe challenges of predicting deformation. Integrating GAT-LSTM with transfer learning technology, remains stable performance whenextended to other tunnels with limited data.
