An extreme learning machine approach for slope stability evaluation and prediction

This paper presents slope stability evaluation and prediction with the approach of a fast robust neural network named the extreme learning machine (ELM). The circular failure mechanism of a slope is formulated based on its material, geometrical and environmental parameters such as the unit weight, the cohesion, the internal friction angle, the slope inclination, slope height and the pore water ratio. The ELM is proposed to evaluate the stability of slopes subjected to potential circular failures by means of prediction of the factor of safety (FS). Substantial slope cases collected worldwide are utilized to illustrate and assess the capability and predictability of the ELM on slope stability analysis. Based on the mean absolute percentage errors and the correlation coefficients between the original and predicted FS values, comparisons are demonstrated between the ELM and the generalized regression neural network (GRNN) as well as the prediction models generated from the genetic algorithms. Moreover, sensitivity analysis of the slope parameters and the ELM model parameters is carried out based on the two utilized evaluation functions. The time expense of the ELM on slope stability analysis is also investigated. The results prove that the ELM is advantageous to the GRNN and the genetic algorithm based models in the analysis of slope stability. Hence, the ELM can be a promising technique for approaching the problems in geotechnical engineering.     

Mapping mineral prospectivity using an extreme learning machine regression

In this research, we conduct a case study of mapping polymetallic prospectivity using an extreme learning machine (ELM) regression. A Quad-Core CPU 1.8 GHz laptop computer served as hardware platform. Almeida's Python program was used to construct the ELM regression model to map polymetallic prospectivity of the Lalingzaohuo district in Qinghai Province in China. Based on geologic, metallogenic, and statistical analyses of the study area, one target and eight predictor map patterns and two training sets were then used to train the ELM regression and logistic regression models. ELM regression modeling using the two training sets spends 61.4 s and 65.9 s; whereas the logistic regression modeling using the two training sets spends 1704.0 s and 1628.0 s. The four trained regression models were used to map polymetallic prospectivity. Based on the polymetallic prospectivity predicted by each model, the receiver operating characteristic (ROC) curve was plotted and the area under the curve (AUC) was estimated. The ROC curves show that the two ELM-regression-based models somewhat dominate the two logistic-regression-based models over the ROC performance space; and the AUC values indicate that the overall performances of the two ELM-regression-based models are somewhat better than those of the two logistic-regression-based models. Hence, the ELM-regression-based models slightly outperform the logistic-regression-based models in mapping polymetallic prospectivity. Polymetallic targets were optimally delineated by using the Youden index to maximize spatial association between the delineated polymetallic targets and the discovered polymetallic deposits. The polymetallic targets predicted by the two ELM-regression-based models occupy lower percentage of the study area (2.66–2.68%) compared to those predicted by the two logistic-regression-based models (4.96%) but contain the same percentage of the discovered polymetallic deposits (82%). Therefore, the ELM regression is a useful fast-learning data-driven model that slightly outperforms the widely used logistic regression model in mapping mineral prospectivity. The case study reveals that the magmatic complexes, which intruded into the Baishahe Formation of the Paleoproterozoic Jinshuikou Group or the Carboniferous Dagangou and Shiguaizi Formations, and which were controlled by northwest-western/east-western trending deep faults, are critical for polymetallic mineralization and need to be paid much attention to in future mineral exploration in the study area.     

Displacement prediction model of landslide based on a modified ensemble empirical mode decomposition and extreme learning machine

In this paper, an M–EEMD–ELM model (modified ensemble empirical mode decomposition (EEMD)-based extreme learning machine (ELM) ensemble learning paradigm) is proposed for landslide displacement prediction. The nonlinear original surface displacement deformation monitoring time series of landslide is first decomposed into a limited number of intrinsic mode functions (IMFs) and one residual series using EEMD technique for a deep insight into the data structure. Then, these sub-series except the high frequency are forecasted, respectively, by establishing appropriate ELM models. At last, the prediction results of the modeled IMFs and residual series are summed to formulate an ensemble forecast for the original landslide displacement series. A case study of Baishuihe landslide in the Three Gorges reservoir area of China is presented to illustrate the capability and merit of our model. Empirical results reveal that the prediction using M–EEMD–ELM model is consistently better than basic artificial neural networks (ANNs) and unmodified EEMD–ELM in terms of the same measurements.     

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

Compression index Ccis an essential parameter in geotechnical design for which the effectiveness of correlation is still a challenge.This paper suggests a novel modelling approach using machine learning(ML)technique.The performance of five commonly used machine learning(ML)algorithms,i.e.back-propagation neural network(BPNN),extreme learning machine(ELM),support vector machine(SVM),random forest(RF)and evolutionary polynomial regression(EPR)in predicting Cc is comprehensively investigated.A database with a total number of 311 datasets including three input variables,i.e.initial void ratio e0,liquid limit water content wL,plasticity index Ip,and one output variable Cc is first established.Genetic algorithm(GA)is used to optimize the hyper-parameters in five ML algorithms,and the average prediction error for the 10-fold cross-validation(CV)sets is set as thefitness function in the GA for enhancing the robustness of ML models.The results indicate that ML models outperform empirical prediction formulations with lower prediction error.RF yields the lowest error followed by BPNN,ELM,EPR and SVM.If the ranges of input variables in the database are large enough,BPNN and RF models are recommended to predict Cc.Furthermore,if the distribution of input variables is continuous,RF model is the best one.Otherwise,EPR model is recommended if the ranges of input variables are small.The predicted correlations between input and output variables using five ML models show great agreement with the physical explanation.     

致密储层渗透率测试的稳态与非稳态法对比研究

渗透率是油气藏储层评价、产能计算及制定合理的开发方案所需关键参数之一。针对常规稳态法测试致密储层岩石渗透率效率低下、试验过程易受环境温度影响等缺点,使用脉冲衰减渗透率测试仪对39块致密岩石稳态法克氏渗透率与非稳态脉冲衰减渗透率进行系统的对比研究,分析岩样渗透率、试验操作方法及有效应力组合方式等因素对测试结果的影响。研究结果表明,相同净围压条件下致密岩石脉冲渗透率小于克氏渗透率,脉冲渗透率平均约为克氏渗透率的47.26%,岩样渗透率越小,则两种渗透率之间的差别越大。误差分析表明,脉冲渗透率试验初期岩样短时间承受的9 MPa高围压和其高围压、高孔隙压力组合的有效应力施加方式对脉冲渗透率测试结果具有一定程度的影响,但仍无法完全解释两种渗透率的总体误差。数学拟合表明,取自同一区块的露头砂岩岩样脉冲与克氏渗透率相对误差与脉冲渗透率具有较好地对数函数关系,由数学推导获得该露头砂岩储层岩石脉冲渗透率与克氏渗透率的转换关系。     

Displacement prediction of step-like landslide by applying a novel kernel extreme learning machine method

Landslide displacement prediction is an essential component for developing landslide early warning systems. In the Three Gorges Reservoir area (TGRA), landslides experience step-like deformations (i.e., periods of stability interrupted by abrupt accelerations) generally from April to September due to the influence of precipitation and reservoir scheduled level variations. With respect to many traditional machine learning techniques, two issues exist relative to displacement prediction, namely the random fluctuation of prediction results and inaccurate prediction when step-like deformations take place. In this study, a novel and original prediction method was proposed by combining the wavelet transform (WT) and particle swarm optimization-kernel extreme learning machine (PSO-KELM) methods, and by considering the landslide causal factors. A typical landslide with a step-like behavior, the Baishuihe landslide in TGRA, was taken as a case study. The cumulated total displacement was decomposed into trend displacement, periodic displacement (controlled by internal geological conditions and external triggering factors respectively), and noise. The displacement items were predicted separately by multi-factor PSO-KELM considering various causal factors, and the total displacement was obtained by summing them up. An accurate prediction was achieved by the proposed method, including the step-like deformation period. The performance of the proposed method was compared with that of the multi-factor extreme learning machine (ELM), support vector regression (SVR), backward propagation neural network (BPNN), and single-factor PSO-KELM. Results show that the PSO-KELM outperforms the other models, and the prediction accuracy can be improved by considering causal factors.     

Prediction of landslide displacement with an ensemble-based extreme learning machine and copula models

Research on the dynamics of landslide displacement forms the basis for landslide hazard prevention. This paper proposes a novel data-driven approach to monitor and predict the landslide displacement. In the first part, autoregressive moving average time series models are constructed to analyze the autocorrelation of landslide triggering factors. A linear ensemble-based extreme learning machine using the least absolute shrinkage and selection operator is applied in predicting the displacement of landslides. Five benchmarking data-driven models, the support vector machine, neural network, random forest, k-nearest neighbor, and the classical extreme learning machine, are considered as baseline models for validating the ensemble-based extreme learning machines. Numerical experiments demonstrated that the proposed prediction model produces the smallest prediction errors among all the algorithms tested. In the second part, parametric copula models are fitted on the predicted displacement, to investigate the relationship between the triggering factors and landslide displacement values. The Gumbel-Hougaard copula model performs best, which indicates strong upper tail correlation between the triggering factors and displacement values. Thresholds for the triggering factors can be obtained by monitoring the landslide moving patterns with large displacement values. The effectiveness and utility of the proposed data-driven approach have been confirmed with the landslide case study in the region of the Three Gorges Reservoir.     

Spatiotemporal modelling of rainfall-induced landslides using machine learning

Landslides - Natural terrain landslides are mainly triggered by rainstorms in Hong Kong, which pose great threats to life and property. To mitigate landslide risk, building a prediction model which...     

A proposal for an approach to mapping susceptibility to landslides using natural language processing and machine learning

Landslides - Compiling an inventory is a fundamental step for carrying out assessments of landslide hazards. However, data in sufficient quantity and quality are not always available. Thus, this...     

Predicting rock size distribution in mine blasting using various novel soft computing models based on meta-heuristics and machine learning algorithms

Blasting is well-known as an effective method for fragmenting or moving rock in open-pit mines. To evaluate the quality of blasting, the size of rock distribution is used as a critical criterion in blasting operations. A high percentage of oversized rocks generated by blasting operations can lead to economic and environmental damage. Therefore, this study proposed four novel intelligent models to predict the size of rock distribution in mine blasting in order to optimize blasting parameters, as well as the efficiency of blasting operation in open mines. Accordingly, a nature-inspired algorithm (i.e., firefly algorithm – FFA) and different machine learning algorithms (i.e., gradient boosting machine (GBM), support vector machine (SVM), Gaussian process (GP), and artificial neural network (ANN)) were combined for this aim, abbreviated as FFA-GBM, FFA-SVM, FFA-GP, and FFA-ANN, respectively. Subsequently, predicted results from the abovementioned models were compared with each other using three statistical indicators (e.g., mean absolute error, root-mean-squared error, and correlation coefficient) and color intensity method. For developing and simulating the size of rock in blasting operations, 136 blasting events with their images were collected and analyzed by the Split-Desktop software. In which, 111 events were randomly selected for the development and optimization of the models. Subsequently, the remaining 25 blasting events were applied to confirm the accuracy of the proposed models. Herein, blast design parameters were regarded as input variables to predict the size of rock in blasting operations. Finally, the obtained results revealed that the FFA is a robust optimization algorithm for estimating rock fragmentation in bench blasting. Among the models developed in this study, FFA-GBM provided the highest accuracy in predicting the size of fragmented rocks. The other techniques (i.e., FFA-SVM, FFA-GP, and FFA-ANN) yielded lower computational stability and efficiency. Hence, the FFA-GBM model can be used as a powerful and precise soft computing tool that can be applied to practical engineering cases aiming to improve the quality of blasting and rock fragmentation.     

A hybrid model using machine learning methods and GIS for potential rockfall source identification from airborne laser scanning data

The main objectives of this paper are to design and evaluate a hybrid approach based on Gaussian mixture model (GMM) and random forest (RF) for detecting rockfall source areas using airborne laser scanning data. The former model was used to calculate automatically slope angle thresholds for different type of landslides such as shallow, translational, rotational, rotational-translational, complex, debris flow, and rockfalls. After calculating the slope angle thresholds, a homogenous morphometric land use area (HMLA) was constructed to improve the performance of the model computations and reduce the sensitivity of the model to the variations in different conditioning factors. After that, the support vector machine (SVM) was applied in addition to backward elimination (BE) to select and rank the conditioning factors considering the type of landslides. Then, different machine learning methods [artificial neural network (ANN), logistic regression (LR), and random forest (RF) were trained with the selected best factors and previously prepared inventory datasets. The best fit method (RF) was then used to generate the probability maps and then the source areas were detected by combining the slope raster (reclassified according to the thresholds found by the GMM model) and the probability maps. The accuracy assessment shows that the proposed hybrid model could detect the potential rockfalls with an accuracy of 0.92 based on training data and 0.96 on validation data. Overall, the proposed model is an efficient model for identifying rockfall source areas in the presence of other types of landslides with an accepted generalization performance.     

Population amount risk assessment of extreme precipitation-induced landslides based on integrated machine learning model and scenario simulation

In this study, the future landslide population amount risk (LPAR) is assessed based on integrated machine learning models (MLMs) and scenario simulation techniques in Shuicheng County, China. Firstly, multiple MLMs were selected and hyperparameters were optimized, and the generated 11 models were cross-integrated to select the best model to calculate landslide susceptibility; by calculating precipitation for different extreme precipitation recurrence periods and combining the susceptibility results to assess the landslide hazard. Using the town as the basic unit, the exposure and vulnerability of the future landslide population under different Shared Socioeconomic Pathways (SSPs) scenarios in each town were assessed, and then combined with the hazard to estimate the LPAR in 2050. The results showed that the integrated model with the optimized random forest model as the combination strategy had the best comprehensive performance in susceptibility assessment. The distribution of hazard classes is similar to susceptibility, and with an increase in precipitation, the low-hazard area and high-hazard decrease and shift to medium-hazard and very high-hazard classes. The high-risk areas for future landslide populations in Shuicheng County are mainly concentrated in the three southwestern towns with high vulnerability, whereas the northern towns of Baohua and Qinglin are at the lowest risk class. The LPAR increased with the intensity of extreme precipitation. The LPAR differs significantly among the SSPs scenarios, with the lowest in the “fossil-fueled development (SSP5)” scenario and the highest in the “regional rivalry (SSP3)” scenario. In summary, the landslide susceptibility model based on integrated machine learning proposed in this study has a high predictive capability. The results of future LPAR assessment can provide theoretical guidance for relevant departments to cope with future socioeconomic development challenges and make corresponding disaster prevention and mitigation plans to prevent landslide risks from a developmental perspective.     

Prediction of Gibbs free energies of calcite-type carbonates and the equilibrium distribution of trace elements between carbonates and aqueous solutions

A linear correlation exists between the standard Gibbs free energies of formation of calcite-type carbonates (MCO3) and the corresponding conventional standard Gibbs free energies of formation of the aqueous divalent cations (M2+) at 25 °C and 1 bar ΔGMCO30 = m(ΔGf,M2+0) ? 141,200 cal · mole?1 where m is equal to 0.9715. This relationship enables prediction of the standard free energies of formation of numerous hypothetical carbonates with the calcite structure. Associated uncertainties typically range from about ± 250 to 600 cal · mole?1. An important consequence of the above correlation is that the thermodynamic equilibrium constant for the distribution of two trace elements M and N between carbonate mineral and aqueous solution at 25 °C and 1 bar is proportional to the free energy difference between the corresponding two aqueous ions: In Combination of predicted standard free energies, entropies and volumes of carbonate minerals at 25°C and 1 bar with standard free energies of aqueous ions and the equation of state in Helgesonet al. (1981) enables prediction of the thermodynamic equilibrium constant for trace element distribution between carbonates and aqueous solutions at elevated temperatures and pressures. Interpretation of the thermodynamic equilibrium constant in terms of concentration ratios in the aqueous phase is considerably simplified if pairs of divalent trace elements are considered that have very similar ionic radii (e.g., $\text{Sr}{}^{\mathrm{2+}}\text{Pb}{}^{\mathrm{2+}}$, $\text{Mg}{}^{\mathrm{2+}}\text{Zn}{}^{\mathrm{2+}}$). In combination with data for the stabilities of complex ions in aqueous solutions, the above calculations enable useful limits to be placed on the concentrations of trace elements in hydrothermal solutions.     

Toward real-time three-dimensional mapping of surficial aquifers using a hybrid modeling approach

A hybrid modeling approach is proposed for near real-time three-dimensional (3D) mapping of surficial aquifers. First, airborne frequency-domain electromagnetic (FDEM) measurements are numerically inverted to obtain subsurface resistivities. Second, a machine-learning (ML) algorithm is trained using the FDEM measurements and inverted resistivity profiles, and borehole geophysical and hydrogeologic data. Third, the trained ML algorithm is used together with independent FDEM measurements to map the spatial distribution of the aquifer system. Efficacy of the hybrid approach is demonstrated for mapping a heterogeneous surficial aquifer and confining unit in northwestern Nebraska, USA. For this case, independent performance testing reveals that aquifer mapping is unbiased with a strong correlation (0.94) among numerically inverted and ML-estimated binary (clay-silt or sand-gravel) layer resistivities (5–20 ohm-m or 21–5,000 ohm-m), and an intermediate correlation (0.74) for heterogeneous (clay, silt, sand, gravel) layer resistivities (5–5,000 ohm-m). Reduced correlation for the heterogeneous model is attributed to over-estimating the under-sampled high-resistivity gravels (about 0.5 % of the training data), and when removed the correlation increases (0.87). Independent analysis of the numerically inverted and ML-estimated resistivities finds that the hybrid procedure preserves both univariate and spatial statistics for each layer. Following training, the algorithms can map 3D surficial aquifers as fast as leveled FDEM measurements are presented to the ML network.     

Order of diagenetic events controls evolution of porosity and permeability in carbonates

The properties of carbonate rocks are often the result of multiple, diagenetic events that involve phases of cementation (porosity occlusion) and dissolution (porosity enhancement). This study tests the hypothesis that the order of these events is a major control on final porosity and permeability. A three-dimensional synthetic model of grainstone is used to quantify trends that show the effect of early cementation, non-fabric selective dissolution, and then a second-generation of (post-dissolution) cement. Models are 3 mm³ with a resolution of 10 μm. Six simple paragenetic sequences are modelled from an identical starting sediment (without accounting for compaction) where the same diagenetic events are placed in different sequences, allowing for quantification of relative changes in the resultant porosity and permeability for each diagenetic event, the trajectory through time, as well as for each final rock. All modelled paragenetic sequences result in reductions in porosity and permeability, but the order of diagenetic events controls the trajectory and final rock properties. Differences in the order of early cement precipitation alone produce variable final values, but all follow the porosity–permeability relationship as expressed by the Kozeny-Carman equation. However, final values for the sequences which include a phase of dissolution fall on a new curve, which departs from that predicted by the Kozeny-Carman relationship. This allows an alternative form of porosity–permeability relationship to be proposed: κ = 2280ϕ–30,400, where ϕ is porosity (%) and κ is permeability (mD). Hence while the Kozeny-Carman relationship predicts porosity–permeability changes that occur with cementation, it is unable to capture accurately changes within the pore network as a result of dissolution. Although the results may be dependent on the properties of the initial carbonate sediment and simplified diagenetic scenarios, it is suggested that this new porosity–permeability relationship may capture some generalized behaviour, which can be tested by modelling further sediment types and diagenetic scenarios.     

Landslide detection in the Himalayas using machine learning algorithms and U-Net

Landslides - Event-based landslide inventories are essential sources to broaden our understanding of the causal relationship between triggering events and the occurring landslides. Moreover,...