We are privileged to be invited by the Honorary Editor-in-Chief,Professor Qihu Qian, Editor-in-Chief, Professor Xia-Ting Feng, and the editorial staff of the Journal of Rock Mechanics and Geotechnical Engineering(JRMGE),to serve as Guest Editors for this Special Issue (SI).
Over the last decade, the application of the Internet of Things(IoT) and Artificial Intelligence (AI) has increased rapidly to enhance automation in various industries. For efficient construction and maintenance of geotechnical infrastructures (slopes, tunnels, pipelines, and other ground infrastructures), there is a need to access and examine measured data in real-time. Variations in data type due to the usage of unmanned aerial vehicle(UAV)photogrammetric sensors, LiDAR, and fiber optic sensing techniques make data management and analysis more complicated.Advanced artificial intelligence,metaheuristic optimization,and data science can be reliable methods in geotechnical engineering for site investigation,risk assessment,design,construction,and maintenance at a higher level.
The purpose of this SI is to review the latest developments in IoT and AI techniques as well as their key applications in solving geotechnical engineering problems. The SI contains 25 papers covering various monitoring methods,machine learning(ML)techniques, and their performance and challenges faced in real world applications. The following is a brief summary of each published paper in this SI.
Liu et al. (2022) adopted seven ML methods for predicting tunneling-induced ground settlement using 14 contributing factors measured before or during tunnel excavation. ML methods were able to successfully predict settlements even though the dataset size was relatively small.
Zhong and Tao (2022) developed a vibrational wireless underground system inspired by subterranean animals that rely on vibrations or seismic waves. The performance of the developed system shows that seismic waves produced by vibration can be used as an information carrier and can potentially be implemented in the Internet of underground things (IoUTs).
Kannangara et al. (2022) compared the application of four different methods,such as Pearson correlation method, sequential forward selection (SFS), sequential backward selection (SBS) and Boruta algorithm, for predicting the tunneling-induced maximum surface settlement. The aim was to understand the influence of feature selection on the predictive ability of these models.
Wu et al. (2022) developed an optical fiber sensing technology to predict the movements of masonry walls and sway of trees in Hong Kong, China. The newly developed system combined the functions of remote sensing and data access, early warning, and real-time data visualization. Major observations include a reversible movement in the masonry retaining wall and the dependence of locations of maximum strain on the crown spread of the trees.
Chen et al.(2022)developed a new method for the characterization of fracture spacing and surface rock quality designation in rock tunnel sections. Their method was validated with field measurements.
Zhang et al. (2022a) found that ensemble-based learning approaches (random forest, extreme gradient boosting) are superior to other methods (support vector machine, logistic regression) for stability prediction of 786 landslide cases in Yunyang County,Chongqing, China. Furthermore, the profile shape was found to be the most significant parameter influencing the stability of slopes.
Zhang et al. (2022b) developed an automatic ML-based approach for modeling excavation-induced tunnel displacements considering the influence of seven input parameters.Two main aspects are investigated,including soil property and spatial characteristics of the deep excavation. The proposed model was validated using field data.
Xiao et al. (2022) adopted the microseismic (MS) monitoring technique in the large-scale underground powerhouse caverns of Shuangjiangkou hydropower station. The progressive fracture damage process of surrounding rock mass was revealed by the spatiotemporal evolution of MS activities.The mechanical response of surrounding rock mass considering MS damage effect was studied through three-dimensional numerical simulation.
Zhu et al.(2022)developed a classification framework to categorize rock blocks of discontinuous rock slopes using a deep convolutional neural network.A total of 1240 high-resolution images from 130 slope masses were used to classify blocks,trapped blocks,and stable blocks based on the principles of Goodman’s block theory.The proposed prediction model was compared with various classifiers and its superior performance was validated.
Xu et al.(2022)developed an intelligent lithology identification method based on deep learning of rock microscopic images. The proposed method is appropriate for geologists to identify lithology in an efficient manner.
Huang et al. (2022) improved existing ML algorithms such as multi-layer perception, support vector machine and gradient boosting regression using a genetic algorithm and principal component analysis for predicting tunnel posture in a shield machine.
Bouzid (2022) utilized the stress deviator increasing method(SDIM) to increase performance of finite element analysis of slope stability. This method allows gradual expansion of the mobilized stress Mohr’s circles until the soil failure according to a prescribed non-convergence criterion. An in-house Fortran computer code called S4DINA was developed for the same and verified by several examples.
Qureshi et al. (2022) utilized multivariate adaptive regression splines to investigate the hydraulic conductivity and discontinuities in the limestone and sandstone formations in Oman.A correlation between hydraulic conductivity and rock quality designation is developed.
Zhang et al.(2022c)developed a Bayesian ML method for probabilistic estimation of the slope failure time, which provides not only a best estimate of the slope failure time, but also the confidence interval of the probable slope failure time. This approach overcomes the limitation of the traditional deterministic inverse velocity method and provides better prediction than the maximum likelihood method.
Du et al.(2022)investigated energy evolution characteristics in the whole process of the disaster chain followed by the establishment of the momentum-conservation equation. The two-phase double point material point method was used to model the landslide-induced surge disaster chain.This method was validated using an experiment involving block-induced surge.
Zhong et al.(2022)explored the deterioration behavior and failure mechanism of rock under the erosion of leaching solution.It is found that soaking time enhances the erosion of quartz crystals inside the rock specimens.The findings from this study are useful for understanding the evolution of damage to basement rock by leaching solution.
Guan et al. (2022) established a smart sampling strategy for obtaining the minimum number of cone penetration tests as well as their optimal locations using information entropy and Bayesian compressive sampling.
Lin et al. (2022) developed a hybrid model based on particle swarm optimization and a gated recurrent unit neural network for tackling time series data of tunnels.The construction variables,such as the main thrust and foam liquid volume,show the highest correlation with the cutterhead torque.
Ganiev et al. (2022) evaluated the effect of short fibers on the mechanical properties of Toyoura sand using a series of consolidated drained triaxial compression tests.The higher the fiber content, the larger the slope of the critical state line at the end of shearing.
Nguyen and Nguyen-Son (2022) developed a cell-based smoothed finite element method for estimating the stability of a square tunnel in the soil.Their proposed method is also applicable in anisotropic and heterogeneous soils.
Cao et al. (2022) conducted a laboratory model test to investigate the strain responses of calcareous sand foundations to variations in the groundwater table and applied loads. During testing,the novel optical frequency domain reflectometry technology with high spatial resolution and accuracy was used to perform distributed strain sensing. The monitoring results indicate that the shear-compression deformation mechanism above the sliding surface was determined by moisture content, loads and particles interlock.The foundation was compressed beneath the sliding surface at all stages.
Cheng et al. (2022) developed an intelligent monitoring platform for the continuous quantification of soil, vegetation, and atmospheric parameters. Models predicting the response time and mean reduction rate of suction as a function of soil depth,initial soil suction, rainfall, tree canopy, air temperature, and wind speed were established using multi-gene genetic programming.
Yan et al. (2022) presented a framework for predicting geological characteristics during earth balance pressure (EPB) shield tunneling based on integrating a stacking classification algorithm with a grid search and K-fold cross validation.
Li et al.(2022)developed a sensing solution for investigating the uplift capacity of pipelines buried in sand using fiber optic strain sensing nerves (FOSS). A series of scaled tests was conducted to analyze the failure mechanism using FOSS and image analyses.
Barman and Dash (2022) critically reviewed various chemical additives for stabilizing expansive soils, such as lime, cement and fly ash. The mechanisms and suitability of these additives under various conditions are discussed.
Although this SI has assembled 26 leading research papers in the field, and provided an opportunity to assess progress to date,the authors would like to propose the future developments and challenges in applications of AI and IoT in geotechnical engineering herein:
(1) Most of the published studies were conducted on limited sets of field data, which have been collected from the literature.There is a lack of systematic big database in terms of geotechnical parameter characterization in the field. The development of a digital database that collects, stores and sorts information on geotechnical parameters is required.This digital database should be made accessible to scientists around the world for further interpretation and to reveal new findings.
(2) There is a need to develop physics based deep learning or ML models for revealing physical mechanisms and capturing trends beyond the existing database. Physics-based ML models integrate data and physics-based models (partial differential equations (PDEs) and mathematical models) to solve certain problems (Vadyala et al., 2022). Physics based learning models need to be developed specifically for geotechnical engineering related problems that can help engineers and scientists simulate certain behaviors of soil as well as capture certain mechanisms in geotechnical engineering.
(3) There is a need to integrate IoT applications in geotechnical engineering with AI techniques for interpreting data and making decisions. Such integration will help in fault diagnosis. An intelligent data acquisition algorithm helps in automatic data collection and processing. It can further reduce the workload of experimentalists and the effect of human factors on the results (Ma and Guo, 2020). Hence,there is a need to develop more advanced IoT-based systems that can automatically collect, process and provide interpretation for quick decision making at both laboratory and field scales.
(4) More applications of UAVs for geotechnical infrastructure monitoring need to be explored.de Sousa Mello et al.(2022)analyzed the topography and settlement of landfills for long term using UAVs.UAV data must be analyzed using combined IoT and AI for further ease of data processing and decision making.
Again, we are grateful to the authors for their generous contributions and patience during the review process. Our heartfelt thanks also go to the dedicated reviewers for their useful comments.
Journal of Rock Mechanics and Geotechnical Engineering2022年4期