This paper aims to establish an intelligent procedure that combines the observational method with the existing deep learning technique for updating deformation of braced excavation in clay.The gated recurrent unit(GRU...This paper aims to establish an intelligent procedure that combines the observational method with the existing deep learning technique for updating deformation of braced excavation in clay.The gated recurrent unit(GRU) neural network is adopted to formulate the forecast model and learn the potential rules in the field observations using the Nesterov-accelerated Adam(Nadam) algorithm.In the proposed procedure,the GRU-based forecast model is first trained based on the field data of previous and current stages.Then,the field data of the current stage are used as input to predict the deformation response of the next stage via the previously trained GRU-based forecast model.This updating process will loop up till the end of the excavation.This procedure has the advantage of directly predicting the deformation response of unexcavated stages based on the monitoring data.The proposed intelligent procedure is verified on two well-documented cases in terms of accuracy and reliability.The results indicate that both wall deflection and ground settlement are accurately predicted as the excavation proceeds.Furthermore,the advantages of the proposed intelligent procedure compared with the Bayesian/o ptimization updating are illustrated.展开更多
1 Introduction For geotechnical engineering,numerous applications involve multiscale and multiphysics processes,such as internal erosion,hydraulic fracturing,energy piles,municipal waste disposal,production from uncon...1 Introduction For geotechnical engineering,numerous applications involve multiscale and multiphysics processes,such as internal erosion,hydraulic fracturing,energy piles,municipal waste disposal,production from unconventional oil and gas reservoirs,heat stimulation and depressurization of natural gas hydrate formation,pavement subjected to heating-cooling cycles.展开更多
The subsoil contains many evaporites such as limestone,gypsum,and salt.Such rocks are very sensitive to water.The deposit of evaporites raises questions because of their dissolution with time and the mechanical-geotec...The subsoil contains many evaporites such as limestone,gypsum,and salt.Such rocks are very sensitive to water.The deposit of evaporites raises questions because of their dissolution with time and the mechanical-geotechnical impact on the neighboring zone.Depending on the configuration of the site and the location of the rocks,the dissolution can lead to surface subsidence and,for instance,the formation of sinkholes and landslides.In this study,we present an approach that describes the dissolution process and its coupling with geotechnical engineering.In the first part we set the physico-mathematical framework,the hypothesis,and the limitations in which the dissolution process is stated.The physical interface between the fluid and the rock(porous)is represented by a diffuse interface of finite thickness.We briefly describe,in the framework of porous media,the steps needed to upscale the microscopic-scale(pore-scale)model to the macroscopic scale(Darcy scale).Although the constructed method has a large range of application,we will restrict it to saline and gypsum rocks.The second part is mainly devoted to the geotechnical consequences of the dissolution of gypsum material.We then analyze the effect of dissolution in the vicinity of a soil dam or slope and the partial dissolution of a gypsum pillar by a thin layer of water.These theoretical examples show the relevance and the potential of the approach in the general framework of geoengineering problems.展开更多
This study proposes an approach that considers mitigation strategies in predicting landslide susceptibility through machine learning(ML)and geographic information system(GIS)techniques.ML models,such as random forest(...This study proposes an approach that considers mitigation strategies in predicting landslide susceptibility through machine learning(ML)and geographic information system(GIS)techniques.ML models,such as random forest(RF),logistic regression(LR),and support vector classification(SVC)are incorporated into GIS to predict landslide susceptibilities in Hong Kong.To consider the effect of mitigation strategies on landslide susceptibility,non-landslide samples were produced in the upgraded area and added to randomly created samples to serve as ML models in training datasets.Two scenarios were created to compare and demonstrate the efficiency of the proposed approach;Scenario I does not considering landslide control while Scenario II considers mitigation strategies for landslide control.The largest landslide susceptibilities are 0.967(from RF),followed by 0.936(from LR)and 0.902(from SVC)in Scenario II;in Scenario I,they are 0.986(from RF),0.955(from LR)and 0.947(from SVC).This proves that the ML models considering mitigation strategies can decrease the current landslide susceptibilities.The comparison between the different ML models shows that RF performed better than LR and SVC,and provides the best prediction of the spatial distribution of landslide susceptibilities.展开更多
To date,the accurate prediction of tunnel boring machine(TBM)performance remains a considerable challenge owing to the complex interactions between the TBM and ground.Using evolutionary polynomial regression(EPR)and r...To date,the accurate prediction of tunnel boring machine(TBM)performance remains a considerable challenge owing to the complex interactions between the TBM and ground.Using evolutionary polynomial regression(EPR)and random forest(RF),this study devel-ops two novel prediction models for TBM performance.Both models can predict the TBM penetration rate and field penetration index as outputs with four input parameters:the uniaxial compressive strength,intact rock brittleness index,distance between planes of weakness,and angle between the tunnel axis and planes of weakness(a).First,the performances of both EPR-and RF-based models are examined by comparison with the conventional numerical regression method(i.e.,multivariate linear regression).Subsequently,the performances of the RF-and EPR-based models are further investigated and compared,including the model robustness for unknown datasets,interior relationships between input and output parameters,and variable importance.The results indicate that the RF-based model has greater prediction accuracy,particularly in identifying outliers,whereas the EPR-based model is more convenient to use by field engineers owing to its explicit expression.Both EPR-and RF-based models can accurately identify the relationships between the input and output param-eters.This ensures their excellent generalization ability and high prediction accuracy on unknown datasets.展开更多
In this paper, we study the dissolution problems occurring in laterally large 3D systems with very small dimensions along the third coordinate, such as fractures or Hele-Shaw cells. On the basis of the scale separatio...In this paper, we study the dissolution problems occurring in laterally large 3D systems with very small dimensions along the third coordinate, such as fractures or Hele-Shaw cells. On the basis of the scale separation assumption, we apply upscaling to the 3D pore-scale model using the volume averaging method to develop 2D averaged equations. The influence of the choice of momentum equations on the accuracy of the 2D Hele-Shaw model is discussed, and we show that the results obtained using Darcy-Brinkman equations are better than those obtained using Darcy’s law, because of the consideration of the viscous boundary layer. The validity and accuracy of the resulting 2D model are assessed based on comparisons with full 3D solutions for problems corresponding to the existence of geometrical 3D features to which a simple averaging procedure along a line(i.e., the height of the gap) perpendicular to the 2D plane cannot be applied, such as the dissolution of pillars. The results show that when Péclet and Reynolds numbers exceed certain limits, 3D effects must be considered. Moreover, natural convection effects are important when the Rayleigh number is large.展开更多
基金The financial supports provided by the Research Grants Council(RGC)of Hong Kong Special Administrative Region Government(HKSARG)of China(Grant Nos.15209119 and PolyU R5037-18F)Zhongtian Construction Group Co.Ltd.(Grant No.ZTCG-GDJTYJSJSFW-2020002)。
文摘This paper aims to establish an intelligent procedure that combines the observational method with the existing deep learning technique for updating deformation of braced excavation in clay.The gated recurrent unit(GRU) neural network is adopted to formulate the forecast model and learn the potential rules in the field observations using the Nesterov-accelerated Adam(Nadam) algorithm.In the proposed procedure,the GRU-based forecast model is first trained based on the field data of previous and current stages.Then,the field data of the current stage are used as input to predict the deformation response of the next stage via the previously trained GRU-based forecast model.This updating process will loop up till the end of the excavation.This procedure has the advantage of directly predicting the deformation response of unexcavated stages based on the monitoring data.The proposed intelligent procedure is verified on two well-documented cases in terms of accuracy and reliability.The results indicate that both wall deflection and ground settlement are accurately predicted as the excavation proceeds.Furthermore,the advantages of the proposed intelligent procedure compared with the Bayesian/o ptimization updating are illustrated.
基金This work is supported by the Research Grants Council(RGC)of the Hong Kong Special Administrative Region Government(HKSARG)of China(No.N_PolyU534/20).
文摘1 Introduction For geotechnical engineering,numerous applications involve multiscale and multiphysics processes,such as internal erosion,hydraulic fracturing,energy piles,municipal waste disposal,production from unconventional oil and gas reservoirs,heat stimulation and depressurization of natural gas hydrate formation,pavement subjected to heating-cooling cycles.
文摘The subsoil contains many evaporites such as limestone,gypsum,and salt.Such rocks are very sensitive to water.The deposit of evaporites raises questions because of their dissolution with time and the mechanical-geotechnical impact on the neighboring zone.Depending on the configuration of the site and the location of the rocks,the dissolution can lead to surface subsidence and,for instance,the formation of sinkholes and landslides.In this study,we present an approach that describes the dissolution process and its coupling with geotechnical engineering.In the first part we set the physico-mathematical framework,the hypothesis,and the limitations in which the dissolution process is stated.The physical interface between the fluid and the rock(porous)is represented by a diffuse interface of finite thickness.We briefly describe,in the framework of porous media,the steps needed to upscale the microscopic-scale(pore-scale)model to the macroscopic scale(Darcy scale).Although the constructed method has a large range of application,we will restrict it to saline and gypsum rocks.The second part is mainly devoted to the geotechnical consequences of the dissolution of gypsum material.We then analyze the effect of dissolution in the vicinity of a soil dam or slope and the partial dissolution of a gypsum pillar by a thin layer of water.These theoretical examples show the relevance and the potential of the approach in the general framework of geoengineering problems.
基金funding by the National Natural Science Foundation of China(Grant No.42007416)the Hong Kong Polytechnic University Strategic Importance Fund(ZE2T)and Project of Research Institute of Land and Space(CD78).
文摘This study proposes an approach that considers mitigation strategies in predicting landslide susceptibility through machine learning(ML)and geographic information system(GIS)techniques.ML models,such as random forest(RF),logistic regression(LR),and support vector classification(SVC)are incorporated into GIS to predict landslide susceptibilities in Hong Kong.To consider the effect of mitigation strategies on landslide susceptibility,non-landslide samples were produced in the upgraded area and added to randomly created samples to serve as ML models in training datasets.Two scenarios were created to compare and demonstrate the efficiency of the proposed approach;Scenario I does not considering landslide control while Scenario II considers mitigation strategies for landslide control.The largest landslide susceptibilities are 0.967(from RF),followed by 0.936(from LR)and 0.902(from SVC)in Scenario II;in Scenario I,they are 0.986(from RF),0.955(from LR)and 0.947(from SVC).This proves that the ML models considering mitigation strategies can decrease the current landslide susceptibilities.The comparison between the different ML models shows that RF performed better than LR and SVC,and provides the best prediction of the spatial distribution of landslide susceptibilities.
基金supported by the research project of Zhongtian Construction Group Co.Ltd.(Grant No.ZTCG-GDJTYJS-JSFW-2020002).
文摘To date,the accurate prediction of tunnel boring machine(TBM)performance remains a considerable challenge owing to the complex interactions between the TBM and ground.Using evolutionary polynomial regression(EPR)and random forest(RF),this study devel-ops two novel prediction models for TBM performance.Both models can predict the TBM penetration rate and field penetration index as outputs with four input parameters:the uniaxial compressive strength,intact rock brittleness index,distance between planes of weakness,and angle between the tunnel axis and planes of weakness(a).First,the performances of both EPR-and RF-based models are examined by comparison with the conventional numerical regression method(i.e.,multivariate linear regression).Subsequently,the performances of the RF-and EPR-based models are further investigated and compared,including the model robustness for unknown datasets,interior relationships between input and output parameters,and variable importance.The results indicate that the RF-based model has greater prediction accuracy,particularly in identifying outliers,whereas the EPR-based model is more convenient to use by field engineers owing to its explicit expression.Both EPR-and RF-based models can accurately identify the relationships between the input and output param-eters.This ensures their excellent generalization ability and high prediction accuracy on unknown datasets.
基金support from the National Natural Science Foundation of China (Grant No. 12102371)Natural Science Foundation of Sichuan Province,China (Grant No. 2022NSFSC1932)。
文摘In this paper, we study the dissolution problems occurring in laterally large 3D systems with very small dimensions along the third coordinate, such as fractures or Hele-Shaw cells. On the basis of the scale separation assumption, we apply upscaling to the 3D pore-scale model using the volume averaging method to develop 2D averaged equations. The influence of the choice of momentum equations on the accuracy of the 2D Hele-Shaw model is discussed, and we show that the results obtained using Darcy-Brinkman equations are better than those obtained using Darcy’s law, because of the consideration of the viscous boundary layer. The validity and accuracy of the resulting 2D model are assessed based on comparisons with full 3D solutions for problems corresponding to the existence of geometrical 3D features to which a simple averaging procedure along a line(i.e., the height of the gap) perpendicular to the 2D plane cannot be applied, such as the dissolution of pillars. The results show that when Péclet and Reynolds numbers exceed certain limits, 3D effects must be considered. Moreover, natural convection effects are important when the Rayleigh number is large.
