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.     

Prediction of rock burst classification using the technique of cloud models with attribution weight

Rock burst is one of the common failures in hard rock mining and civil construction. This study focuses on the prediction of rock burst classification with case instances using cloud models and attribution weight. First, cloud models are introduced briefly related to the rock burst classification problem. Then, the attribution weight method is presented to quantify the contribution of each rock burst indicator for classification. The approach is implemented to predict the classes of rock burst intensity for the 164 rock burst instances collected. The clustering figures are generated by cloud models for each rock burst class. The computed weight values of the indicators show that the stress ratio $ Ts = \sigma_{\theta } /\sigma_{c} $ Ts = σ θ / σ c is the most vulnerable parameter and the elastic strain energy storage index W _et and the brittleness factor $ B = \sigma_{c} /\sigma_{t} $ B = σ c / σ t take the second and third place, respectively, contributing to the rock burst classification. Besides, the predictive performance of the strategy introduced in this study is compared with that of some empirical methods, the regression analysis, the neural networks and support vector machines. The results turn out that cloud models perform better than the empirical methods and regression analysis and have superior generalization ability than the neural networks in modelling the rock burst cases. Hence, cloud models are feasible and applicable for prediction of rock burst classification. Finally, different models with varying indicators are investigated to validate the parameter sensitivity results obtained by cloud clustering analysis and regression analysis in context to rock burst classification.     

Numerical study on the dynamic behavior of rock avalanche: influence of cluster shape,size and gradation

Acta Geotechnica - In this paper, we present a numerical study of dynamic behavior of rock avalanche by using three-dimensional particle flow simulation. The emphasis is put on the influence of...     

Comprehensive Stability Evaluation of Rock Slope Using the Cloud Model-Based Approach

This article presents the cloud model-based approach for comprehensive stability evaluation of complicated rock slopes of hydroelectric stations in mountainous area. This approach is based on membership cloud models which can account for randomness and fuzziness in slope stability evaluation. The slope stability is affected by various factors and each of which is ranked into five grades. The ranking factors are sorted into four categories. The ranking system of slope stability is introduced and then the membership cloud models are applied to analyze each ranking factor for generating cloud memberships. Afterwards, the obtained cloud memberships are synthesized with the factor weights given by experts for comprehensive stability evaluation of rock slopes. The proposed approach is used for the stability evaluation of the left abutment slope in Jinping 1 Hydropower Station. It is shown that the cloud model-based strategy can well consider the effects of each ranking factor and therefore is feasible and reliable for comprehensive stability evaluation of rock slopes.     

Effects of confining pressure and loading path on deformation and strength of cohesive granular materials: a three-dimensional DEM analysis

This paper is devoted to numerical analysis of strength and deformation of cohesive granular materials. The emphasis is put on the study of effects of confining pressure and loading path. To this end, the three-dimensional discrete element method is used. A nonlinear failure criterion for inter-granular interface bonding is proposed, and it is able to account for both tensile and shear failure for a large range of normal stress. This criterion is implemented in the particles flow code. The proposed failure model is calibrated from triaxial compression tests performed on representative sandstone. Numerical results are in good agreement with experimental data. In particular, the effect of confining pressure on compressive strength and failure pattern is well described by the proposed model. Furthermore, numerical predictions are studied, respectively, for compression and extension tests with a constant mean stress. It is shown that the failure strength and deformation process are clearly affected by loading path. Finally, a series of numerical simulations are performed on cubic samples with three independent principal stresses. It is found that the strength and failure mode are strongly influenced by the intermediate principal stress.

     

Comparison on landslide nonlinear displacement analysis and prediction with computational intelligence approaches

Landslide displacement is widely obtained to discover landslide behaviors for purpose of event forecasting. This article aims to present a comparative study on landslide nonlinear displacement analysis and prediction using computational intelligence techniques. Three state-of-art techniques, the support vector machine (SVM), the relevance vector machine (RVM), and the Gaussian process (GP), are comparatively presented briefly for modeling landslide displacement series. The three techniques are discussed comparatively for both fitting and predicting the landslide displacement series. Two landslides, the Baishuihe colluvial landslide in China Three Georges and the Super-Sauze mudslide in the French Alps, are illustrated. The results prove that the computational intelligence approaches are feasible and capable of fitting and predicting landslide nonlinear displacement. The Gaussian process, on the whole, performs better than the support vector machine, relevance vector machine, and simple artificial neural network (ANN) with optimized parameter values in predictive analysis of the landslide displacement.