Display omitted
•A new hybrid SMBO-CatBoost algorithm that integrates CatBoost with the Sequential Model-Based Optimization is developed.•The present model predicts with more accuracy than the ...optimized XGBoost, SVM, RF, KNN, LR, and AdaBoost.•Unnecessary input variables are removed to improve the models’ field applicability.•A new method for more accurately calculating the rock mass boreability is proposed based on the big data.
In-time perception of changing geological conditions is crucial for safe and efficient TBM tunneling. Precisely detecting or predicting the rock mass qualities ahead of the tunnel face can forewarn the geological disasters (e.g., burst or squeezing behaviors of surrounding rock mass). A novel hybridization model based on CatBooost and Sequential Model-Based Optimization (SMBO) is proposed in this study. Firstly, a database incorporating 4464 samples acquired from the Songhua River Water Diversion Project is established using the capping method. Owing to SMBO’s different surrogate types (GP, RF, and GBRT) and performance validation, the comparisons of SMBO-CatBoost’s three types and other six hybridized models (SMBO-XGBoost, SMBO-AdaBoost, SMBO-RF, SMBO-SVM, SMBO-KNN, and SMBO-LR) are successively carried out. As a result, in terms of the optimization speed, performance, and sensitivity to poor geological conditions, SMBO(RF)-CatBoost is the most suitable model for rock mass class prediction; furthermore, it achieves the best performance ACC¯ = 0.9207 and F1¯ = 0.9178 among the seven hybridized models. Next, the scientific feature selection methods (i.e., filter, embedded) are used to reduce the model’s complexity (i.e., feature dimensions) step by step to increase the model’s on-site practicality. The determined ten influential features still can keep the model’s ACC¯ and F1¯ greater than 0.85, and only respectively declines 5.4% and 5.6% in contrast to the original performance. Subsequently, in order to explore the importance of the first-hand features and the second-hand features (i.e., composite features), a new method for more accurately calculating the rock mass boreability indices (regarded as the second-hand features) is proposed based on the big data at a relatively high sampling frequency of 1 Hz, this newly-proposed method could make these indices more of significance under the complex geological conditions. With the SHAP technique, the modified torque penetration index (TPI’) is more valuable than other second-hand and some first-hand features.
A CHN-BQ method is firstly defined to present the system and connotation of the current Chinese national standard “Standards for Engineering Classification of Rock Mass” GB/T 50,218. The core idea ...and content of the CHN-BQ method are studied in detail. Based on a qualitative classification of rock masses at the sites of in situ rock deformation and shear strength tests with a total number of 1439 test sets in nearly 90 projects, the statistical characteristics of in situ test results of different grades of rock mass are analyzed; Thus, the mechanical parameters corresponding to each grade of rock mass basic quality are proposed. Based on 997 groups of various rock mass classification data in 10 projects, the comprehensive CHN-BQ method verifies, and some essential remarks are obtained. Firstly, the 312 sets of parallelly measured data show that the consistency of less than one-grade discrepancy of the qualitative and quantitative classification in the CHN-BQ method is up to 71% in percentage. Secondly, a highly linear correlativity of the CHN-BQ method with the RMR method is found, and a linear regression equation is proposed. Thirdly, a logarithmic regression equation between the CHN-BQ method and the Q system is obtained, and equivalent grading divisions of the Q system from 9-level intervals towards traditional 5-level intervals are proposed. Finally, the research shows that the GSI concept in the Hoek-Brown criterion and the qualitative classification of rock mass structure in the CHN-BQ method are essentially coincident. Based on the measured data in quartz mica schist rock of an arch dam project, a linear regression equation of the Kv index in the CHN-BQ method and GSI index in the Hoek-Brown criterion is obtained, and the comparison of the strength parameters of the CHN-BQ method and Hoek-Brown criterion is discussed.
The Hoek–Brown criterion was introduced in 1980 to provide input for the design of underground excavations in rock. The criterion now incorporates both intact rock and discontinuities, such as ...joints, characterized by the geological strength index (GSI), into a system designed to estimate the mechanical behaviour of typical rock masses encountered in tunnels, slopes and foundations. The strength and deformation properties of intact rock, derived from laboratory tests, are reduced based on the properties of discontinuities in the rock mass. The nonlinear Hoek–Brown criterion for rock masses is widely accepted and has been applied in many projects around the world. While, in general, it has been found to provide satisfactory estimates, there are several questions on the limits of its applicability and on the inaccuracies related to the quality of the input data. This paper introduces relatively few fundamental changes, but it does discuss many of the issues of utilization and presents case histories to demonstrate practical applications of the criterion and the GSI system.
•An ensemble model AdaBoost-CART is proposed to predict rock mass classification.•SMOTE is utilized to address the imbalance ratio of rock mass classifications.•The AdaBoost-CART model performs ...better than conventional machine learning methods.•The variable importance of the model is analyzed.
The real-time acquisition of surrounding rock information is important for the efficient tunneling and hazard prevention in tunnel boring machines (TBMs). This study presents an ensemble learning model based on classification and regression tree (CART) and AdaBoost algorithm to predict the classification of surrounding rock mass. Statistical indicators (i.e., mean value and standard deviation) of TBM operational parameters were calculated and used as input variables, and the rock mass classification obtained by the hydropower classification (HC) method was used as output variable. To develop the model, a database was established, consisting of 3166 samples collected from the Songhua River Water Conveyance Tunnel. The synthetic minority over-sampling technique (SMOTE) was utilized to address the imbalance ratio of rock mass classifications in the database. The results of the testing set showed that the accuracy and F1-measure of AdaBoost-CART were 0.865 and 0.770, respectively, which are better than the results of the standard CART (0.753 and 0.629, respectively). The application of SMOTE improves the recall of minority classes. Compared with artificial neural networks, k-nearest neighbor, and support vector classifier, the developed AdaBoost-CART model achieves better performance. The variable importance was analyzed to distinguish key features; the results showed that rock mass boreability may not be a major consideration of the HC method. The presented model can provide significant guidance for the real-time acquisition of surrounding rock information during TBM tunneling.
The strength and deformability of rock masses transected by non-persistent joints are controlled by complex interactions of joints and intact rock bridges. The emergence of synthetic rock mass (SRM) ...numerical modelling offers a promising approach to the analysis of rock masses, but has not been rigorously compared with actual physical experiments. In this work, SRM modelling by the discrete element software PFC3D is used to investigate the effect of geometric parameters of joints on the rock mass failure mechanism, unconfined compressive strength and deformation modulus. Firstly, a validation study is undertaken to investigate the ability of SRM modelling to reproduce rock mass failure modes and strength as determined by uniaxial and biaxial compression testing in the laboratory. The numerical analyses agree well with physical experimentation at low confining pressure. A sensitivity study is then undertaken of the effect of joint configuration parameters on the failure mode, unconfined compressive strength and deformation modulus of the rock mass. Five failure modes are predicted to occur: intact rock, planar, block rotation, step-path and semi-block generation. It is found that the failure mode is determined principally by joint orientation and step angle and the joint orientation with respect to principal stress direction is the parameter with the greatest influence on rock mass properties.
A numerical characterization of a fractured rock mass and its mechanical behavior using a discontinuum approach was carried out utilizing lattice-spring-based synthetic rock mass (LS-SRM) models. ...First, LS-SRM models on a laboratory scale were created to reproduce standard rock mechanical tests on Triassic sandstone samples from a quarry in Germany. Subsequently, the intact rock properties were upscaled to an element volume representative for geotechnical applications, recalibrated and combined with a Discrete Fracture Network (DFN) model. The resulting fractured rock mass properties are compared to predictions from empirical relationships based on rock mass classification schemes and the DFN-Oda-Geomechanics approach. Modeling results reveal a significant reduction in the strength of the fractured rock mass compared to the intact rock, showing a high agreement with empirically calculated values. Results for the deformation modulus reveal a significant reduction induced by the fracture network and a good agreement compared to the results obtained by other approaches. It is shown that the LS-SRM allows analyzing the complex mechanical behavior during failure of rock masses, including crack initiation, propagation and coalescence. The resulting rock mass properties are key parameters for a wide range of geotechnical applications and can be used for large-scale numerical modeling as well.
The rock mass classification methods developed to evaluate rock mass stability and tunnel support design for blast and drill tunnels are not suitable to guide tunneling by tunnel boring machine ...(TBM), such as selection of TBM types, determination of construction scheme and so on. By comprehensively considering the rock mass boreability and stability, a modified rock mass classification system for TBM tunnels is proposed on the basis of the hydropower rock mass classification (HC) method of China. The input parameters are five rock mass parameters as the same as the HC method. In addition, the tunnel diameter is needed to estimate the TBM advance rate. The structure and parameter determination of the classification system are explained and statistically analyzed in details. This system can be applied to evaluate the feasibility of TBM construction and serve as the basis for selection of TBM types and design parameters in the pre-construction phase. Furthermore, it can also be applied to estimate TBM performance and guide tunnel support for a given ground condition in the design phase, and to optimize TBM operational parameters during the construction phase as well. An example is presented to illustrate the application of the proposed system. The limitations of the proposed system and further studies are also discussed.
•A modified rock mass classification system for TBM tunnels and tunneling is developed.•The system has the characteristics of simple input parameters and structure.•The system can estimate the rock mass stability, boreability and TBM applicability.•The database is established with a wide range of rock types and rock mass parameters.
This paper describes synthetic rock mass (SRM) modeling, a new approach for simulating the mechanical behavior of jointed rock mass. This technique uses the bonded particle model for rock to ...represent intact material and the smooth-joint contact model (SJM) to represent the in situ joint network. The macroscopic behavior of an SRM sample depends on both the creation of new fractures through intact material and slip/opening of pre-existing joints. SRM samples containing thousands of non-persistent joints can be submitted to standard laboratory tests (UCS, triaxial loading, and direct tension tests) or tested under a non-trivial stress path representative of the stresses induced during the engineering activity under study.
Output from the SRM methodology includes pre-peak properties (modulus, damage threshold, peak strength, etc.) and post-peak properties (brittleness, dilation angle, residual strength, fragmentation, etc.). Of particular interest is the ability to obtain predictions of rock mass scale effects, anisotropy, and brittleness, properties that cannot be obtained using empirical methods of property estimation. This paper presents the theoretical background of the SRM approach along with some example applications.
Rock mass classification is essential for assessing the quality of macroscopic rock mass and is the basis for rock mass stability analysis and geotechnical engineering design. The joint observation ...technology limits traditional rock mass classification methods in that they only collect joint information from one-dimensional or two-dimensional space and cannot comprehensively obtain the joint occurrence in three-dimensional space. Consequently, empirical formulas are frequently used in studies on joint distribution laws, resulting in less accurate calculations of joint parameters. This study develops a method for classifying rock masses using a precise description of the joints. Initially, it utilizes the borehole camera and the Sirovision joint scanning system to acquire accurate three-dimensional joint occurrence data. The subjective and the objective weights of each evaluation index are derived from the analytic hierarchy process (AHP) and the CRITIC technique according to the cloud model theory. The game theory is then employed to determine the combined weight and evaluate the quality of a rock mass method with the cloud model (GA-CM). The proposed classification method is applied to the slope of an open-pit mine. The results indicate that compared to the traditional methods, the proposed method is objective, accurate, and field-applicable and also reduces the influence of subjective factors on rock mass quality evaluation and enhances the classification reliability.
Current assessments of rock mass quality of a NATM tunnel face are important in the practice of tunnel excavation. This study establishes a multi-source database and proposes a data driven method for ...the assessment. Thirteen multi-source variables describing the tunnel faces are considered as inputs, and the rock mass rating (RMR) values computed by the empirical formula are the target outputs. We adopted two meta machine learning models (classification and regression tree (CART) and multiple layers perceptron (MLP)) and two ensemble learning models (gradient boosting regression tree (GBRT)) and random forest (RF)) to capture the relationships between the inputs and outputs. The tree-structured Parzen estimator (TPE) algorithm is applied to automatically determine the optimized model hyper-parameters. The experimental results suggest that the proposed hybrid ensemble learning models (TPE-RF and TPE-GBRT) perform well at assessing rock mass quality. The feature importance ranks of the input variables are determined by a sensitivity analysis, which enhances the knowledge on assessing the rock mass quality of a tunnel face.
Display omitted
PDF
