Xinyu Wang’s research while affiliated with Wuhan University and other places

Uniaxial compressive strength (UCS) is a critical parameter in the disaster prevention of engineering projects, requiring a large budget and a long time to estimate in different rocks or the early stage of a project. If predicted accurately, the UCS of rocks significantly affects geotechnical applications. This paper develops a dataset of 734 samples from previous studies on different coun-tries' magmatic, sedimentary, and metamorphic rocks. Within the study context, three main factors, point load index, P-wave velocity, and Schmidt hammer rebound number, are utilized to estimate UCS. Moreover, it applies extreme learning machines (ELM) to map the nonlinear relationship between the UCS and the influential factors. Five metaheuristic algorithms, particle swarm optimization (PSO), grey wolf optimization (GWO), whale optimization algorithm (WOA), butterfly optimization algorithm (BOA), and sparrow search algorithm (SSA), are used to optimize the bias and weight of ELM and thus enhance its predictability. Indeed, several performance parameters are utilized to verify the proposed models' generalization capability and predictive performance. The minimum, maximum, and average relative errors of ELM achieved by the whale optimization algorithm (WOA-ELM) are smaller than the other models, with values of 0.22%, 72.05%, and 11.48%, respectively. In contrast, the minimum and mean residual error produced by WOA-ELM are less than the other models, with values of 0.02 and 2.64 MPa, respectively. The results show that the UCS values derived from WOA-ELM are superior to those from other models. The performance indices (coefficient of determination (R 2): 0.861, mean squared error (MSE): 17.61, root mean squared error (RMSE): 4.20, and value account for (VAF): 91% obtained using the WOA-ELM model indicates high accuracy and reliability, which means that it has broad application potential for estimating UCS of different rocks.

Cutting force is an essential parameter of the operating state of tunneling boring machines. In this study, a real-time cutting force monitoring method is proposed for widely used disc-cutters. A microcylindrical strain gauge is embedded in predrilled small boreholes in the bottom and side surface of the two C-shaped cushion blocks on the cutter saddle. The strain gauges are protected by a glue seal, and the sensor leads are protected using a groove and covered by wear-resistant steel. The strain is measured using a wireless strain node and transmitted from the strain node to the wireless gateway connected to a computer. Then, the strain-force relationship of the cushion block is fitted using a laboratory static calibration test. The load on the cushion block is calculated based on the measured cushion block strain. Subsequently, considering the cutter system’s vibration, a multi-degree-of-freedom coupling vibration differential equation of the cutter system is established. The mass matrix, stiffness matrix, and damping matrix of the cutter system are also determined. Finally, a Wilson-θ inverse analysis method is put forward to calculate the cutter’s external load. The developed real-time cutting force monitoring method will not affect cutter replacement, and the cushion block equipped with sensors can be reused. The cutting force identification method, which considers the cutter system vibration, is rigorous in theory, and a useful numerical inverse analysis calculation method is provided.

In order to determine the operational status of tunnel boring machines (TBMs), a real-time cutting tool force monitoring system is established based on the proposed cutting tool force measurement and calculation method. The monitoring system is composed of strain sensors, a data acquisition module, a data transmission module, and a data analysis module. The sensor layout and the complete monitoring system operating principle are presented. A dynamic rotary rock cutting test was carried out to verify the effectiveness of the proposed cutting tool force measurement method and corresponding monitoring system. Finally, the monitoring system was applied to the ‘Colorful Clouds’-named TBM in the Gaoligongshan Tunnel (China). This monitoring system can provide continuous power supply, unattended automatic operation, and monitor error information feedback, thereby realizing real-time monitoring, remote control, and data sharing. The field monitoring results show that the cutter's maximum peak normal force and maximum peak side force appear every few seconds. Every two normal force peaks mark a complete penetration process. The maximum and minimum normal force peaks are approximately 800 kN and 50 kN, respectively, and the maximum normal force peak is approximately 6–7 times the peak side force. The normal force frequency indicates that the cutter's normal force is concentrated from 0 to 300 kN, accounting for approximately 95% of the overall proportion.

The stability of the shafts connecting underground mineral resources with the surface can directly affect the efficiency and operational safety of mining operations. In this study, factors affecting shaft stability are analyzed, and eight evaluation indexes are summarized. Compared with single weighting methods, a combination weighting method based on the sum of squared deviations can improve the accuracy of weighting the evaluation indexes. First, the indexes are subjectively weighted using an analytic hierarchy process (AHP), and the objective weights are calculated according to the criteria importance through intercriteria correlation (CRITIC). Then, the combined weights are calculated based on the maximum sum of squared deviations. The preference ranking organization method for enrichment evaluation (PROMETHEE) II decision-making algorithm is then used to transform the shaft stability evaluation into a multi-attribute group decision-making problem. The algorithm is verified by seven coal mine shaft cases in eastern China, and the algorithm achieves an accuracy of 100%. Furthermore, the influence of the Gaussian preference function, linear preference function, and 0-1 preference function on the performance of PROMETHEE II decision-making algorithm is analyzed. The Gaussian preference function is shown to be superior.

Coal androck identification is one of the key technologies to realize intelligent and unmanned mining and heading. In order to further improve the accuracy and efficiency of coal rock image recognition technology, a coal-rock image segmentation model(Coal-Rock Pyramid Network, CRPN) based on spatial pyramid pooling architecture and convolutional neural network is proposed in this paper. ① Depthwise separable convolution is used in the coding part of CRPN. With the support of hybrid dilated convolution and residual convolution embedded with the global attention mechanism, the computational efficiency and the receptive field have been improved, and the adverse effects of global irrelevant features on subsequent feature maps are reduced. The computing framework based on the spatial pooling architecture is applied to the decoding part, weakening the loss of related information between different regions in the feature map, and enhancing the effective representation of global information. ② To ensure the effectiveness of model training, The in-situ coal and rock images from the thin seam mining face are collected by high sensitivity underground explosion-proof SLR camera, including four typical conditions:complete state, crack and shadow, dark light, dark light with wire mesh supporting. The image preprocessing was executed, including noise addition and changing the characteristics and morphology of the image. A high-definition coal-rock image database containing 6 400 valid samples has been established. ③ A training optimization algorithm based on cross-entropy loss function and rectified adaptive moment estimation is proposed, taking into account the efficiency and accuracy. ④ The pixel accuracy and intersection over union are selected as indicators to evaluate the recognition effect of CRPN. The results show that the average values of the two indicators of CRPN are 96.05% and 91.54%, respectively, which are better than other existing coal-rock recognition models such as U-net and Segnet. The average calculation time of a single image of CRPN model is 0. 037 s, which is higher than the frame rate of mining explosion-proof camera equipment(25 fps) and proves its application deployment ability. The CRPN model was deployed in the dynamic video obtained on the working face for testing. The test results show that the model has achieved correct coal-rock recognition results under both stable and jitter conditions, verifying its feasibility and robustness in complex environments. © 2021, Editorial Office of Journal of China Coal Society. All right reserved.

Rockburst is a common dynamic geological hazard, severely restricting the development and utilization of underground space and resources. As the depth of excavation and mining increases, rockburst tends to occur frequently. Hence, it is necessary to carry out a study on rockburst prediction. Due to the nonlinear relationship between rockburst and its influencing factors, artificial intelligence was introduced. However, the collected data were typically imbalanced. Single algorithms trained by such data have low recognition for minority classes. In order to handle the problem, this paper employed stacking technique of ensemble learning to establish rockburst prediction models. In total, 246 sets of data were collected. In the preprocessing stage, three data mining techniques including principal component analysis, local outlier factor and expectation maximization algorithm were used for dimension reduction, outlier detection and outlier substitution, respectively. Then, the pre-processed data were split into a training set (75%) and a test set (25%) with stratified sampling. Based on the four classical single intelligent algorithms, namely k-nearest neighbors (KNN), support vector machine (SVM), deep neural network (DNN) and recurrent neural network (RNN), four ensemble models (KNN-RNN, SVM-RNN, DNN-RNN and KNN-SVM-DNN-RNN) were built by stacking technique of ensemble learning. The prediction performance of eight models was evaluated, and the differences between single models and ensemble models were analyzed. Additionally, a sensitivity analysis was conducted , revealing the importance of input variables on the models. Finally, the impact of class imbalance on the prediction accuracy and fitting effect of models was quantitatively discussed. The results showed that stacking technique of ensemble learning provides a new and promising way for rockburst prediction, which exhibits unique advantages especially when using imbalanced data.

Rockburst is a common dynamic geological hazard, severely restricting the development and utilization of underground space and resources. As the depth of excavation and mining increases, rockburst tends to occur frequently. Hence, it is necessary to carry out a study on rockburst prediction. Due to the nonlinear relationship between rockburst and its influencing factors, artificial intelligence was introduced. However, the collected data were typically imbalanced. Single algorithms trained by such data have low recognition for minority classes. In order to handle the problem, this paper employed stacking technique of ensemble learning to establish rockburst prediction models. In total, 246 sets of data were collected. In the preprocessing stage, three data mining techniques including principal component analysis, local outlier factor and expectation maximization algorithm were used for dimension reduction, outlier detection and outlier substitution, respectively. Then, the pre-processed data were split into a training set (75%) and a test set (25%) with stratified sampling. Based on the four classical single intelligent algorithms, namely k-nearest neighbors (KNN), support vector machine (SVM), deep neural network (DNN) and recurrent neural network (RNN), four ensemble models (KNN-RNN, SVM-RNN, DNN-RNN and KNN-SVM-DNN-RNN) were built by stacking technique of ensemble learning. The prediction performance of eight models was evaluated, and the differences between single models and ensemble models were analyzed. Additionally, a sensitivity analysis was conducted , revealing the importance of input variables on the models. Finally, the impact of class imbalance on the prediction accuracy and fitting effect of models was quantitatively discussed. The results showed that stacking technique of ensemble learning provides a new and promising way for rockburst prediction, which exhibits unique advantages especially when using imbalanced data.

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.

Rockburst is a kind of dynamic instability phenomenon that easily occurs in the excavation process of deep underground engineering. It has the characteristics of suddenness, uncertainty and strong destructiveness. The research on the classification and prediction of rockburst intensity has become a worldwide problem that needs urgent solution. Due to the classification and prediction of rockburst intensity is a typical multi-attribute ordered segmentation problem, the prediction model is established by using attribute interval recognition theory. Considering the causes of rockburst and its characteristics，the ratio of the maximum tangential stress of surrounding rock to the uniaxial compressive strength of rock, the ratio of the uniaxial compressive strength to the tensile strength of rock, the elastic strain energy index, and the intactness index of rock mass are selected as the evaluation indexes from the aspects of physical and mechanical properties of rock，rock mass integrity and in-situ stress. The subjective weight and objective weight of these evaluation indexes are determined by the analytic hierarchy process ( AHP) and anti-entropy weight method respectively, overcoming the problem that the traditional entropy weight method is sensitive to the index difference when determining the objective weight, and based on which, an optimal combined weighting rule on the basis of the sum of squares of deviations is proposed. Further, the optimal combined weighting-attribute interval recognition model for the classification and prediction of rockburst intensity is established. 12 groups of typical rockburst engineering cases are chosen to test the proposed model. Since the averaging coefficient has a great influence on the prediction performance of the model, in order to select the optimal averaging coefficient, it is changed within the interval [0.05, 0.95] and the step size is 0.1. After the analysis, it is found that when the averaging coefficient is 0.05 and 0.15, the prediction accuracy of the model is the highest, reaching 91.7%. Finally, the prediction result with averaging coefficient being 0.15 is showed and compared with the fuzzy comprehensive evaluation method, the grey evaluation model and the actual situations, which indicates that the proposed model in this paper is feasible and applicable.