A spatiotemporal deep learning method for excavation-induced wall deflections

作     者：Yuanqin Tao Shaoxiang Zeng Honglei Sun Yuanqiang Cai Jinzhang Zhang Xiaodong Pan Yuanqin Tao;Shaoxiang Zeng;Honglei Sun;Yuanqiang Cai;Jinzhang Zhang;Xiaodong Pan

作者机构：College of Civil EngineeringZhejiang University of TechnologyHangzhou310023China Ministry of Education(MOE)Key Laboratory of Soft Soils and Geoenvironmental Engineering(SSGeo)Zhejiang UniversityHangzhou310058China Key Laboratory of Geotechnical and Underground Engineering of Ministry of Education and Department of Geotechnical EngineeringTongji UniversityShanghai200092China 

出 版 物：《Journal of Rock Mechanics and Geotechnical Engineering》 (岩石力学与岩土工程学报（英文版）)

年 卷 期：2024年第16卷第8期

页      面：3327-3338页

核心收录：

学科分类：08[工学] 080104[工学-工程力学] 0815[工学-水利工程] 0801[工学-力学（可授工学、理学学位）] 

基　　金：supported by the National Natural Science Foundation of China(Grant No.42307218) the Foundation of Key Laboratory of Soft Soils and Geoenvironmental Engineering(Zhejiang University),Ministry of Education(Grant No.2022P08) the Natural Science Foundation of Zhejiang Province(Grant No.LTZ21E080001) 

主　　题：Braced excavation Wall deflections Deep learning Convolutional layer Long short-term memory(LSTM) Sequence to sequence(seq2seq) 

摘      要：Data-driven approaches such as neural networks are increasingly used for deep excavations due to the growing amount of available monitoring data in practical ***,most neural network models only use the data from a single monitoring point and neglect the spatial relationships between multiple monitoring ***,most models lack flexibility in providing predictions for multiple days after monitoring *** study proposes a sequence-to-sequence(seq2seq)two-dimensional(2D)convolutional long short-term memory neural network(S2SCL2D)for predicting the spatiotemporal wall deflections induced by deep *** model utilizes the data from all monitoring points on the entire wall and extracts spatiotemporal features from data by combining the 2D convolutional layers and long short-term memory(LSTM)*** S2SCL2D model achieves a long-term prediction of wall deflections through a recursive seq2seq *** excavation depth,which has a significant impact on wall deflections,is also considered using a feature fusion *** excavation project in Hangzhou,China,is used to illustrate the proposed *** results demonstrate that the S2SCL2D model has superior prediction accuracy and robustness than that of the LSTM and S2SCL1D(one-dimensional)*** prediction model demonstrates a strong generalizability when applied to an adjacent *** on the long-term prediction results,practitioners can plan and allocate resources in advance to address the potential engineering issues.