Search

This study employs a deep learning approach to determine the resilient modulus of compacted subgrade, which is one of the most important stiffness characteristics in pavement design. The proposed paradigm, i.e., deep neural network (DNN), comes under the category of artificial neural network with several hidden layers and activation functions. A total of 2813 data of subgrade soils, comprising six influencing parameters namely weighted plasticity index, dry unit weight, confining stress, deviator stress, moisture content, and the number of freezing-thaw cycles, were considered for the creation and validation of the model. The results of the employed DNN were compared with those of other benchmark techniques, such as feed-forward neural network, k-nearest neighbour regressor, extreme learning machine, random forests regressor, multivariate adaptive regression spline, and multiple linear regression. As per the determination coefficient (R2) and root mean square error (RMSE) indices, the developed DNN achieved the maximum degree of precision of robust modulus during both training (R2= 0.9947 and RMSE = 0.0094) and testing (R2= 0.9797 and RMSE = 0.0183) phases. The study also employed DNN-based monotonicity analysis to examine the effects of different influencing parameters. Overall, the developed DNN has demonstrated the potential to assist geotechnical and geological engineers in estimating the resilient modulus of compacted subgrade at varying freezing-thaw cycles and moistures during the preliminary phase of the engineering projects. The developed Python code is attached for future research.

This study aims to circumvent the operation of conducting laboratory tests of soil shear strength through a hybrid machine learning approach. The proposed approach integrates extreme learning machine (ELM) and particle swarm optimisation (PSO) with adaptive acceleration coefficients. Three hybrid ELMs, namely PSO optimised ELM with time-varying acceleration coefficients (ELM-TP), ELM optimised by improved PSO (ELM-IP), and ELM optimised by modified PSO (ELM-MP), have been established. Subsequently, the concept of mean PSO has been incorporated, and three additional hybrid models, namely ELM-TP integrated with mean PSO (ELM-TMP), ELM-IP integrated with mean PSO (ELM-IMP), and ELM-MP integrated with mean PSO (ELM-MMP), are constructed. The proposed concept is also used to construct six artificial neural network (ANN)-based hybrid models (i.e. ANN-TP, ANN-IP, ANN-MP, ANN-TMP, ANN-IMP, and ANN-MMP). Experimental results exhibit that the constructed ELM-IMP and ANN-IMP models can achieve the most desired accuracies in predicting the shear strength of soils.

Tight carbonate reservoirs appear to be heterogeneous due to the patchy production of various digenetic properties. Consequently, the permeability calculation of tight rocks is costly, and only a finite number of core plugs in any single reservoir can be estimated. Hence, in the present study, a novel hybrid model constructed by combination of the improved version of the Harris Hawks optimisation (HHO), i.e., IHHO, and extreme learning machine (ELM) is proposed to predict the permeability of tight carbonates using limited number of input variables. The proposed IHHO employs a mutation mechanism to avoid trapping in local optima by increasing the search capabilities. Subsequently, ELM-IHHO, a novel metaheuristic ELM-based algorithm, was developed to predict the permeability of tight carbonates. Experimental results show that the proposed ELM-IHHO attained the most accurate prediction with R2= 0.9254 and RMSE = 0.0619 in the testing phase. The result of the proposed model is significantly better than those obtained from other ELM-based hybrid models developed with particle swarm optimisation, genetic algorithm, and slime mould algorithm. The results also illustrate that the proposed ELM-IHHO model outperforms the other benchmark model, such as back-propagation neural nets, support vector regression, random forest, and group method of data handling in predicting the permeability of tight carbonates.

Thermal conductivity is a specific thermal property of soil which controls the exchange of thermal energy. If predicted accurately, the thermal conductivity of soil has a significant effect on geothermal applications. Since the thermal conductivity is influenced by multiple variables including soil type and mineralogy, dry density, and water content, its precise prediction becomes a challenging problem. In this study, novel computational approaches including hybridisation of two metaheuristic optimisation algorithms, i.e. firefly algorithm (FF) and improved firefly algorithm (IFF), with conventional machine learning techniques including extreme learning machine (ELM), adaptive neuro-fuzzy interface system (ANFIS) and artificial neural network (ANN), are proposed to predict the thermal conductivity of unsaturated soils. FF and IFF are used to optimise the internal parameters of the ELM, ANFIS and ANN. These six hybrid models are applied to the dataset of 257 soil cases considering six influential variables for predicting the thermal conductivity of unsaturated soils. Several performance parameters are used to verify the predictive performance and generalisation capability of the developed hybrid models. The obtained results from the computational process confirmed that ELM-IFF attained the best predictive performance with a coefficient of determination = 0.9615, variance account for = 96.06%, root mean square error = 0.0428, and mean absolute error = 0.0316 on the testing dataset (validation phase). The results of the models are also visualised and analysed through different approaches using Taylor diagrams, regression error characteristic curves and area under curve scores, rank analysis and a novel method called accuracy matrix. It was found that all the proposed hybrid models have a great ability to be considered as alternatives for empirical relevant models. The developed ELM-IFF model can be employed in the initial stages of any engineering projects for fast determination of thermal conductivity.

The behavior of rock masses is influenced by a variety of forces, with measurement of stress and strain playing the most critical roles in assessing deformation. The laboratory test for determining strain at each location within rock samples is expensive and difficult but rock strain data are important for predicting failure of rock material. Many researchers employ AI technology in order to solve these difficulties. AI algorithms such as gradient boosting machine (GBM), support vector regression (SVR), random forest (RF), and group method of data handling (GMDH) are used to efficiently estimate the strain at every point within a rock sample. Additionally, the ensemble unit (EnU) may be utilized to evaluate rock strain. In this study, 3000 experimental data are used for the purpose of prediction. The obtained strain values are then evaluated using various statistical parameters and compared to each other using EnU. Ranking analysis, stress-strain curve, Young’s modulus, Poisson’s ratio, actual vs. predicted curve, error matrix and the Akaike’s information criterion (AIC) values are used for comparing models. The GBM model achieved 98.16% and 99.98% prediction accuracy (in terms of values of R2) in the longitudinal and lateral dimensions, respectively, during the testing phase. The GBM model, based on the experimental data, has the potential to be a new option for engineers to use when assessing rock strain.

Slope failures lead to catastrophic consequences in numerous countries and thus the stability assessment for slopes is of high interest in geotechnical and geological engineering researches. A hybrid stacking ensemble approach is proposed in this study for enhancing the prediction of slope stability. In the hybrid stacking ensemble approach, we used an artificial bee colony (ABC) algorithm to find out the best combination of base classifiers (level 0) and determined a suitable meta-classifier (level 1) from a pool of 11 individual optimized machine learning (OML) algorithms. Finite element analysis (FEA) was conducted in order to form the synthetic database for the training stage (150 cases) of the proposed model while 107 real field slope cases were used for the testing stage. The results by the hybrid stacking ensemble approach were then compared with that obtained by the 11 individual OML methods using confusion matrix, F1-score, and area under the curve, i.e. AUC-score. The comparisons showed that a significant improvement in the prediction ability of slope stability has been achieved by the hybrid stacking ensemble (AUC = 90.4%), which is 7% higher than the best of the 11 individual OML methods (AUC = 82.9%). Then, a further comparison was undertaken between the hybrid stacking ensemble method and basic ensemble classifier on slope stability prediction. The results showed a prominent performance of the hybrid stacking ensemble method over the basic ensemble method. Finally, the importance of the variables for slope stability was studied using linear vector quantization (LVQ) method.

The feasibility of preconditioned conjugate gradient (PCG) for solving extremely large sparse linear systems arising in three-dimensional finite element limit analysis of geotechnical stability problems is studied. With problem size increasing beyond few millions of variables, the use of direct solvers becomes prohibitive (both in storage and computational times) even on modern computers. This motivates the study on possible employment of iterative methods to solve systems of equations resulting from interior point method-based limit analysis formulations. In this paper, an adaptive preconditioning technique based on incomplete Cholesky (IC) preconditioner, as well as a novel implementation scheme for preconditioned conjugate gradient method, is proposed to effectively deal with highly ill-conditioned systems, such as those which occur at last iterations of IPM methods. Numerical tests on some ultra-large scale sample geotechnical problems, along with comparison with other potential solution methods, verify the robustness and efficiency of the proposed approach.

The study proposes an improved Harris hawks optimization (IHHO) algorithm by integrating the standard Harris hawks optimization (HHO) algorithm and mutation-based search mechanism for developing a high-performance machine learning solution for predicting soil compression index. HHO is a newly introduced meta-heuristic optimization algorithm (MOA) used to solve continuous search problems. Compared to the original HHO, the proposed IHHO can evade trapping in local optima, which in turn raises the search capabilities and enhances the search mechanism relying on mutation. Subsequently, a novel meta-heuristic-based soft computing technique called ELM-IHHO was established by integrating IHHO and extreme learning machine (ELM) to estimate soil compression index. A sum of 688 consolidation test data was collected for this purpose from an ongoing dedicated freight corridor railway project. To evaluate the generalization capability of the proposed ELM-IHHO model, a detailed comparison between ELM-IHHO and other well-established MOAs, such as particle swarm optimization, genetic algorithm, and biogeography-based optimization integrated with ELM, was performed. Based on the outcomes, the ELM-IHHO model exhibits superior performance over the other MOAs in predicting soil compression index.

This study implements a hybrid ensemble machine learning method for forecasting the rate of penetration (ROP) of tunnel boring machine (TBM), which is becoming a prerequisite for reliable cost assessment and project scheduling in tunnelling and underground projects in a rock environment. For this purpose, a sum of 185 datasets was collected from the literature and used to predict the ROP of TBM. Initially, the main dataset was utilised to construct and validate four conventional soft computing (CSC) models, i.e. minimax probability machine regression, relevance vector machine, extreme learning machine, and functional network. Consequently, the estimated outputs of CSC models were united and trained using an artificial neural network (ANN) to construct a hybrid ensemble model (HENSM). The outcomes of the proposed HENSM are superior to other CSC models employed in this study. Based on the experimental results (training RMSE = 0.0283 and testing RMSE = 0.0418), the newly proposed HENSM is potential to assist engineers in predicting ROP of TBM in the design phase of tunnelling and underground projects.