期刊：
Geotechnical and Geological Engineering,2024年42(8):7385-7405 ISSN：0960-3182
通讯作者：
Da Hu
作者机构：
[Shurong Feng] Hunan Provincial Key Laboratory of Key Technology on Hydropower Development, PowerChina Zhongnan Engineering Corporation Limited, Changsha, China;[Yongjia Hu; Rong Hu] School of Civil Engineering, Hunan City University, Yiyang, China;Hunan Engineering Research Centre for Structural Safety and Disaster Prevention of Urban Underground Infrastructure, Hunan City University, Yiyang, China;[Da Hu; Yongsuo Li; Ze Tan] Hunan Engineering Research Centre for Structural Safety and Disaster Prevention of Urban Underground Infrastructure, Hunan City University, Yiyang, China<&wdkj&>School of Civil Engineering, Hunan City University, Yiyang, China
通讯机构：
[Da Hu] H;Hunan Engineering Research Centre for Structural Safety and Disaster Prevention of Urban Underground Infrastructure, Hunan City University, Yiyang, China<&wdkj&>School of Civil Engineering, Hunan City University, Yiyang, China
关键词：
Intelligent classification of surrounding rock;Stability of surrounding rock;Deep learning;Extreme learning machine;ISSA-ELM combined model
摘要：
During construction, mountainous highway tunnels are often subjected to complex forces and are prone to large deformations, which severely affect the long-term stability of the surrounding rock. Therefore, it is crucial to explore rapid identification and intelligent classification methods for rock surrounding tunnels. In response to the above issues, this study proposes a new intelligent classification and prediction method for rock surrounding highway tunnels on the basis of an index classification system of the environmental characteristics of rock surrounding highway tunnels combined with deep learning algorithms. This method can optimize the generative adversarial network for tabular data (CTGAN) via a genetic algorithm (GA) to increase the data volume and then use the Kolmogorov–Smirnov (K–S) test to determine the optimal parameters and samples in the CTGAN with a small number of samples. By combining the sparse search algorithm (SSA) and extreme learning machine (ELM) to construct the SSA-ELM model, the Singer mapping method is used to handle the sparrow random initialization problem, the parameters of the SSA-ELM model are further optimized via the K-fold cross-validation method, and the ISSA-ELM combination model is established. Finally, on the basis of actual engineering cases, 160 sets of data were optimized to analyse and evaluate the classification of surrounding rocks, verifying the rationality and effectiveness of the model. The research results show that the ISSA-ELM combination model minimizes the negative impact of subjective factors on the model and has the advantages of extremely low error, accurate stability, and high robustness. This can provide an important reference for predicting the stability of the surrounding rock in mountain road tunnels.
期刊：
INTERNATIONAL JOURNAL OF COMPUTATIONAL METHODS,2024年 ISSN：0219-8762
通讯作者：
Ni, Pengpeng;Chen, Y
作者机构：
[Xiang, Xuejuan; Tan, Ze; Liu, Jing; Hu, Rong; Hu, Da; Hu, Yongjia; Li, Yongsuo] Hunan City Univ, Hunan Engn Res Ctr Struct Safety & Disaster Preven, Yiyang 413000, Peoples R China.;[Hu, Da; Li, Yongsuo] Hunan City Univ, Sch Civil Engn, Yiyang 413000, Peoples R China.;[Ni, Pengpeng; Ni, PP; Chen, Yu] Sun Yat Sen Univ, Sch Civil Engn, Guangzhou 510275, Peoples R China.;[Ni, Pengpeng; Ni, PP; Chen, Yu] State Key Lab Tunnel Engn, Guangzhou 510275, Peoples R China.;[Ni, Pengpeng; Ni, PP; Chen, Yu] Southern Marine Sci & Engn Guangdong Lab Zhuhai, Zhuhai 519082, Peoples R China.
通讯机构：
[Ni, PP; Chen, Y ] S;Sun Yat Sen Univ, Sch Civil Engn, Guangzhou 510275, Peoples R China.;State Key Lab Tunnel Engn, Guangzhou 510275, Peoples R China.;Southern Marine Sci & Engn Guangdong Lab Zhuhai, Zhuhai 519082, Peoples R China.
关键词：
Settlement prediction;finite element simulation;intelligent optimization algorithm;mesh optimization;finite element-machine learning coupling model
摘要：
Ground settlement prediction for shield construction is highly important and challenging. This study introduces a machine learning algorithm combined with finite element numerical simulation, i.e., machine learning-finite element mesh optimization. For surface subsidence prediction, 16 combination models of ANN, KNN, RF and SVR were optimized by PSO, GA, BT and BO, involving raw data preprocessing, principal component analysis, hyperparameter selection and prediction accuracy evaluation. A subway shield tunneling project was analyzed, in which the meshes of finite element numerical models were discretized into different sizes from 1.0m to 2.0m. In total, 360 sets of data points were extracted from the simulation results, including stress, strain, shield jacking force, internal friction angle, cohesion force, and settlement, of which 252 data points were used as the input parameters of machine learning model. Analysis of average error rate of finite element-machine learning coupling models showed that the finite element model had the highest accuracy of settlement prediction when the mesh size of the finite element model was 1.4m, and the GA-SVR model had the highest accuracy and generalization ability in ground settlement prediction. This study highlights the uniqueness of machine learning-finite element mesh optimization model in application.
作者机构：
[Liang, Xiaoqiang; Yang, Xian; Hu, Da; Li, Yongsuo; Hu, D] Hunan City Univ, Hunan Engn Res Ctr Struct Safety & Disaster Preven, Yiyang 413000, Peoples R China.;[Yi, Shun; Liang, Xiaoqiang; Yang, Xian; Hu, Da; Hu, Yongjia; Hu, D] Hunan City Univ, Coll Civil Engn, Yiyang 413000, Peoples R China.;[Hu, Da; Hu, D] Power China Zhongnan Engn Co Ltd, Hunan Prov Key Lab Key Technol Hydropower Dev, Changsha 410014, Peoples R China.
通讯机构：
[Hu, D ] H;Hunan City Univ, Hunan Engn Res Ctr Struct Safety & Disaster Preven, Yiyang 413000, Peoples R China.;Hunan City Univ, Coll Civil Engn, Yiyang 413000, Peoples R China.;Power China Zhongnan Engn Co Ltd, Hunan Prov Key Lab Key Technol Hydropower Dev, Changsha 410014, Peoples R China.
摘要：
To provide theoretical support for the safety control of rectangular pipe jacking tunnels crossing an existing expressway, a method for predicting the surface settlement of a rectangular pipe jacking tunnel is proposed in this study. Therefore, based on the high approximation of the BP neural network to any function under the multiparameter input, the PSO-BP mixed prediction model of the ground subsidence of the ultrashallow buried large section rectangular pipe jacking tunnel is established by taking into account the adaptive mutation method, adopting the improved particle swarm optimization (IPSO) algorithm with adaptive inertia weight and mutation particles in the later stage to determine the optimal hyperparameters of the prediction model. Through the case study of an ultrashallow large cross-section rectangular pipe jacking tunnel, this algorithm is compared with the traditional algorithm and combined with field monitoring data for analysis and prediction. The prediction results show that compared with the traditional BP neural network prediction model, AWPSO-BP model and PWPSO-BP model, the improved PSO-BP mixed prediction model shows a more stable prediction effect when the change in surface subsidence is gentle and the concavity and convexity are large. The predicted subsidence value is close to the actual value, and the accuracy and robustness of the prediction are significantly improved.
作者机构：
[姚琦; 冯涛; 廖泽] Hunan Provincial Key Laboratory of Safe Mining Techniques of Coal Mines, Hunan University of Science and Technology, Xiangtan;Hunan;411201, China;[黎永索] School of Civil Engineering, Hunan City University, Yiyang;413000, China
通讯机构：
[Feng, T.] H;Hunan Provincial Key Laboratory of Safe Mining Techniques of Coal Mines, China
关键词：
sandstone and mudstone particles;rheological deformation;segmented rheological model;rheological limit strain;rheological test
摘要：
A mixture of sandstone and mudstone particles is often used as the main filling material for many agriculture-related and civil engineering projects, including rock-fill dams and foundations. The long-term rheological deformation experienced by rock-fill dams and foundations filled with this mixture is much bigger than that of coarse-grained and cohesive soils, due to the deterioration and softening of the mudstone particles. This study focuses on the rheological deformation of a sandstone-mudstone particle mixture, prepared by mixing sandstone and mudstone particles, based on the content by weight of four mudstone particle types. Confined uniaxial compression tests were performed to test the rheological deformation of 24 samples of the mixture, and a stress-strain curve was obtained for each test. On the basis of compression curves, the rheological process of the mixture was divided into four phases: linear, attenuation rheological, secondary attenuation rheological and stable phases. The three defining features of the curve, namely the rheological attenuation factors, attenuation rheology critical strain and limited rheological strain, were then determined and modeled. A segmented rheological model was then proposed, based on a modified attenuation rheological constitutive model for coarse-grained soil. The modelled results compared well with the experimental data, and the modelled compression-curve prediction was able to describe the two-stage attenuation rheology features (attenuation rheological and secondary attenuation rheological phases) of the sandstone-mudstone particle mixture.
期刊：
Geotechnical and Geological Engineering,2017年35(4):1439-1451 ISSN：0960-3182
通讯作者：
Guo, Youlin(guoyoulin82@163.com)
作者机构：
[Guo, Youlin; Zhao, Minghua] Geotechnical Engineering Institute, Hunan University, Changsha;410082, China;[Fu, Guihai; Li, Yongsuo] College of Civil Engineering, Hunan City University, Yiyang;413000, China;[Yu, Pengfei] Yiyang City Highway Administration Bureau of Hunan Province, Yiyang
通讯机构：
[Youlin Guo] G;Geotechnical Engineering Institute, Hunan University, Changsha, China<&wdkj&>College of Civil Engineering, Hunan City University, Yiyang, China
