Review article
Comprehensive review of machine learning in geotechnical reliability analysis: Algorithms, applications and further challenges

https://doi.org/10.1016/j.asoc.2023.110066Get rights and content

Highlights

  • A state-of-the-art review of ML in geotechnical reliability analysis applications.

  • Several commonly used ML algorithms and some latest advanced ML methods are summarized.

  • The potential challenges and prospective research possibilities are outlined.

Abstract

Geotechnical reliability analysis provides a novel way to rationally take the underlying geotechnical uncertainties into account and evaluate the stability of geotechnical structures by failure probability (or equivalently, reliability index) from a probabilistic perspective, which has gained great attention in the past few decades. With the rapid development of artificial intelligence techniques, various machine learning (ML) algorithms have been successfully applied in geotechnical reliability analysis and the number of relevant papers has been increasing at an accelerating pace. Although significant advances have been made in the past two decades, a systematic summary of this subject is still lacking. To better conclude current achievements and further shed light on future research, this paper aims to provide a state-of-the-art review of ML in geotechnical reliability analysis applications. Through reviewing the papers published in the period from 2002 to 2022 with the topic of applying ML in the reliability analysis of slopes, tunneling, and excavations, the pros and cons of the developed methods are explicitly tabulated. The great achievements that have been made are systematically summarized from two major aspects. In addition, the four potential challenges and prospective research possibilities underlying geotechnical reliability analysis are also outlined, including multisensor data fusion, time-variant reliability analysis, three-dimensional reliability analysis of practical cases, and ML model selection and optimization.

Introduction

The safety of geotechnical structures has gained increasing attention in geotechnical engineering practice because their failure may cause significant casualties and economic losses, such as the catastrophic Qianjiangping landslide (e.g., [1]) and Shenzhen landslide [2]. Evaluating the stability of geotechnical structures reasonably is an important prerequisite for disaster prevention and reduction, and geotechnical reliability analysis provides a novel way to explicitly consider the underlying various geotechnical uncertainties (e.g., the inherent variability of geomaterial properties, transformation uncertainty, and measurement error) (e.g., [3], [4]) and quantify the safety margin of geotechnical structures by failure probability (or equivalently, reliability index) from a probabilistic perspective. Generally, the evaluation of failure probability is not a trivial task in geotechnical reliability analysis since the complicated implicit performance functions are commonly encountered in engineering practice, leading to the unavailability of analytical solutions. In such a case, the brute-force Monte Carlo simulation (MCS) method has gained popularity in reliability analysis by virtue of its simplicity, flexibility, and easy to use for geotechnical engineers. Nonetheless, it suffers from a known criticism of extensive computational efforts and poor efficiency (e.g., [5], [6], [7]). The MCS needs to repeatedly invoke geotechnical software (e.g., Abaqus, FLAC, and GeoStudio) to perform a large number of deterministic analyses so as to reach the desired accuracy, which may be a computationally expensive task in practical applications.

With the rapid development of artificial intelligence technologies, many researchers have contributed to the integration of ML and geotechnical reliability analysis for improving computational accuracy and efficiency, which gives rise to a lot of successful applications (e.g., [5], [8], [9], [10], [11], [12], [13], [14], [15], [16]). The basic idea of ML in aiding geotechnical reliability analysis is to reconstruct the high-dimensional implicit performance function through learning from the prepared data which mainly contains the input random variables or random field samples of geomaterial properties (e.g., cohesion, friction angle, and saturated hydraulic conductivity) as well as the corresponding quantity of interest (e.g., the factor of safety) that is generally calculated from the geotechnical software. As the ML-based reliability analysis model reaches the desired performance after sufficient training and proper validation, it can be conveniently used to estimate the failure probability of geotechnical structures with reasonable accuracy and efficiency.

Until now, various ML algorithms and their variants have been successfully applied in the geotechnical reliability analysis, such as artificial neural network (ANN) (e.g., [17]), support vector machine (SVM) (e.g., [18]), relevant vector machine (RVM) (e.g., [15]), particle swarm optimization (PSO) (e.g., [19]), extreme learning machine (ELM) (e.g., [20]), multivariate adaptive regression spline (MARS) (e.g., [10], [21]), extreme gradient boosting (XGBoost) (e.g., [11], [22], [23]), and convolutional neural network (CNN) [7], [13], [14], [24], [25]. These ML algorithms greatly enhance the computational efficiency in the geotechnical reliability analysis, allowing engineers and researchers to focus more on the engineering problems without being compromised by the prohibitively computational tasks in practical applications. Furthermore, benefiting from the great development of ML, the free and open-source packages of the above-mentioned ML algorithms are widely available today from the Internet, such as the open-source packages of the XGBoost and CNN are shared in the well-known GitHub website and can be downloaded for users freely. This open and friendly environment has greatly accelerated the development of ML-based geotechnical reliability analysis methods. Through retrieving from the Web of Science database, a total of 306 papers have been published in the period from 2002 to 2022 concerned with applying ML in the reliability analysis of slopes, tunneling, and excavations. In the past two decades, more and more geotechnical practitioners have been devoted to the application of ML in geotechnical reliability analysis, and this research topic is expected to have great prospects.

Although significant advances have been made in geotechnical reliability analysis in the past two decades, a systematic summary on this subject is still lacking. To the best of the authors’ knowledge, this paper is the first comprehensive survey of recent progress in ML-based geotechnical reliability analysis. The main contributions of this review can be summarized as follows: (1) It systematically reviews 306 papers with the topic of applying ML in the geotechnical reliability analysis of slopes, tunneling, and deep excavations, and the pros and cons of the developed methods are explicitly tabulated using three tables; (2) It provides insights into the previous research from two important aspects, namely the input geotechnical parameter characterization and output failure probability evaluation in geotechnical reliability analysis; (3) It outlines four crucial but unsolved problems in geotechnical reliability analysis, as well as novel challenges and promising research directions.

This paper aims to present a review of ML in geotechnical reliability analysis applications. The remainder of this paper starts with a brief introduction of failure probability evaluation in Section 2. In Section 3, several commonly used ML algorithms and their applications in the geotechnical reliability analysis are systematically outlined. Besides, the insights from previous works are also outlined in Section 4. In Section 5, the potential challenges and prospective research possibilities are presented. Finally, the primary conclusions drawn from this review are summarized in Section 6.

Section snippets

Failure probability evaluation

In geotechnical reliability analysis, the failure probability Pf (or reliability index β) is frequently used to measure the safety margin of geotechnical structures from a probabilistic perspective, which can be defined as (e.g., [4]): Pf=P[g(Θ)<0]=g(Θ)<0f(Θ)dΘwhere Θ denotes geotechnical parameters that are commonly regarded as random variables or random field variables; g(Θ) is the performance function concerning geotechnical parameters Θ; P[g(Θ)<0] is the possibility of g(Θ) less than

Applications of ML in geotechnical reliability analysis

From the statistical data retrieved from the Web of Science database, the main applications of ML in the reliability analysis of geotechnical structures focus on slopes, tunneling, and excavations. Fig. 1 depicts the proportion of ML applications in these three fields. It is obvious that most ML applications are devoted to the slope (i.e., 48%) in the past two decades, followed by excavation, and tunneling accounts for the smallest proportion. If ‘machine learning’, ‘slope’, and ‘reliability’

Insights from previous works

After reviewing the research contributed by previous researchers, great achievements have been made in geotechnical reliability analysis with the aid of ML, which are summarized from two important aspects in this section.

Challenges and future directions

Benefited from the great development of AI technologies, more and more ML algorithms including several commonly used ML algorithms and some latest advanced ML methods (e.g., XGBoost and CNN) have been applied to the geotechnical reliability analysis in the past decades, which gives rise to a lot of ML-based reliability analysis methods and greatly facilitates the implementation of reliability analysis in geotechnical engineering practice. Although great achievements have been obtained, there

Summary and conclusions

This paper reviewed previous studies on the applications of ML in different geotechnical reliability analysis problems in the past two decades. The evaluation of failure probability is a primary concern in the geotechnical reliability analysis and the direct MCS method usually requires extensive computational efforts to ensure the desired accuracy. With the great development of ML in the past few decades, many researchers have contributed to the evaluation of failure probability by

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgments

The authors are grateful to the financial supports from National Natural Science Foundation of China (52008058 and 52078086), Program of Distinguished Young Scholars, Natural Science Foundation of Chongqing, China (cstc2020jcyj-jq0087), and High-end Foreign Expert Introduction program, China (DL2021165001L, G20200022005, and G2022165004L). Special thanks are given to the anonymous reviewers for their valuable comments and constructive suggestions.

References (84)

  • YiP. et al.

    Cumulative PSO-Kriging model for slope reliability analysis

    Probab. Eng. Mech.

    (2015)
  • ShenH. et al.

    Prediction of load–displacement performance of grouted anchors in weathered granites using FastICA-MARS as a novel model

    Geosci. Front.

    (2021)
  • ZhaoT. et al.

    Interpolation and stratification of multilayer soil property profile from sparse measurements using machine learning methods

    Eng. Geol.

    (2020)
  • LiJ. et al.

    Advanced prediction of tunnel boring machine performance based on big data

    Geosci. Front.

    (2021)
  • LiuZ. et al.

    Modelling of shallow landslides with machine learning algorithms

    Geosci. Front.

    (2021)
  • ZhangP. et al.

    Intelligent modelling of clay compressibility using hybrid meta-heuristic and machine learning algorithms

    Geosci. Front.

    (2021)
  • ZhangR.H. et al.

    Estimation of diaphragm wall deflections for deep braced excavation in anisotropic clays using ensemble learning

    Geosci. Front.

    (2021)
  • ZhengS. et al.

    Probabilistic outlier detection for sparse multivariate geotechnical site investigation data using Bayesian learning

    Geosci. Front.

    (2021)
  • Karkevandi-TalkhoonchehA. et al.

    Modeling minimum miscibility pressure during pure and impure CO2 flooding using hybrid of radial basis function neural network and evolutionary techniques

    Fuel

    (2018)
  • ZhangW.G. et al.

    Multivariate adaptive regression splines for analysis of geotechnical engineering systems

    Comput. Geotech.

    (2013)
  • ZhangW.G. et al.

    Multivariate adaptive regression splines and neural network models for prediction of pile drivability

    Geosci. Front.

    (2016)
  • MoH. et al.

    Developing window behavior models for residential buildings using XGBoost algorithm

    Energy Build.

    (2019)
  • KangF. et al.

    Slope reliability analysis using surrogate models via new support vector machines with swarm intelligence

    Appl. Math. Model.

    (2016)
  • ZhaoH. et al.

    Reliability analysis of tunnel using least square support vector machine

    Tunn. Undergr. Space Technol.

    (2014)
  • BucherC. et al.

    A comparison of approximate response functions in structural reliability analysis

    Probab. Eng. Mech.

    (2008)
  • LiX. et al.

    A hybrid approach combining uniform design and support vector machine to probabilistic tunnel stability assessment

    Struct. Saf.

    (2016)
  • BaiY.C. et al.

    Comparative study of metamodeling techniques for reliability analysis using evidence theory

    Adv. Eng. Softw.

    (2012)
  • WangQ. et al.

    Reliability analysis of tunnels using a metamodeling technique based on augmented radial basis functions

    Tunn. Undergr. Space Technol.

    (2016)
  • PanQ. et al.

    An efficient reliability method combining adaptive support vector machine and Monte Carlo simulation

    Struct. Saf.

    (2017)
  • CuiF. et al.

    Implementation of machine learning techniques into the subset simulation method

    Struct. Saf.

    (2019)
  • SamuiP. et al.

    Utilization relevance vector machine for slope reliability analysis

    Appl. Soft Comput.

    (2011)
  • HamrouniA. et al.

    Probability analysis of shallow circular tunnels in homogeneous soil using the surface response methodology optimized by a genetic algorithm

    Tunn. Undergr. Space Technol.

    (2019)
  • MoreiraN. et al.

    Back analysis of geomechanical parameters in underground works using an evolution strategy algorithm

    Tunn. Undergr. Space Technol.

    (2013)
  • ZhuangD.Y. et al.

    Mechanical parameter inversion in tunnel engineering using support vector regression optimized by multi-strategy artificial fish swarm algorithm

    Tunn. Undergr. Space Technol.

    (2019)
  • ZhangJ. et al.

    Deep learning-based evaluation of factor of safety with confidence interval for tunnel deformation in spatially variable soil

    J. Rock Mech. Geotech. Eng.

    (2021)
  • ZhangC. et al.

    Application of optimized grey discrete Verhulst-BP neural network model in settlement prediction of foundation pit

    Environ. Earth Sci.

    (2019)
  • ZhangW. et al.

    State-of-the-art review of soft computing applications in underground excavations

    Geosci. Front.

    (2020)
  • LeungY.F. et al.

    Quantifying cost-effectiveness of subsurface strata exploration in excavation projects through geostatistics and spatial tessellation

    Autom. Constr.

    (2018)
  • WangY. et al.

    Simulation of random fields with trend from sparse measurements without detrending

    J. Eng. Mech.

    (2019)
  • ZhangJ.Z. et al.

    Novel approach to estimate vertical scale of fluctuation based on CPT data using convolutional neural networks

    Eng. Geol.

    (2021)
  • HuangD. et al.

    Towards a complete understanding of the triggering mechanism of a large reactivated landslide in the Three Gorges reservoir

    Eng. Geol.

    (2018)
  • WangF.W. et al.

    Mechanism for the rapid motion of the Qianjiangping landslide during reactivation by the first impoundment of the Three Gorges dam reservoir

    China. Landslides

    (2008)
  • Cited by (34)

    • A spatiotemporal deep learning method for excavation-induced wall deflections

      2024, Journal of Rock Mechanics and Geotechnical Engineering
    View all citing articles on Scopus
    View full text