Tunnelling and Underground Space Technology最新文献

{"title":"Coupled thermo-mechanical simulation of lining cracking evolution and sealing system mechanical response in CAES lined rock caverns using finite-discrete element method","authors":"Zhangxing Wang ,&nbsp;Jiao Wang ,&nbsp;Guanhua Sun ,&nbsp;Shan Lin ,&nbsp;Zhijun Liu ,&nbsp;Hong Zheng","doi":"10.1016/j.tust.2026.107460","DOIUrl":"10.1016/j.tust.2026.107460","url":null,"abstract":"<div><div>Lined rock caverns (LRCs) have become a key underground solution for large-scale compressed air energy storage (CAES). Clarifying the lining’s cracking pattern is a prerequisite for achieving coordinated performance with the sealing layer. This study proposes a coupled thermo-mechanical numerical framework based on the finite-discrete element method, which can effectively predict the random cracking process and crack evolution patterns of the lining. The accuracy and applicability of the proposed framework are verified through comparison with results from laboratory model tests. Finally, an engineering-scale model is constructed to investigate the effects of factors such as thermal effects, surrounding rock stiffness, and reinforcement parameters on the cracking characteristics and mechanical performance of the lining-sealing system. Results show that thermally induced circumferential compression offsets tensile stresses caused by internal pressure, reducing the maximum crack width and the steel liner stress amplitude by approximately 30%. Surrounding rock stiffness governs deformation compatibility: a higher elastic modulus suppresses plastic zone expansion, significantly reduces cracking, and improves the stress uniformity of the steel liner. Reinforcement factors (including reinforcement type, bar diameter, and spacing) have a limited effect on crack development and overall stress in the steel liner but influence the uniformity of stress in the sealing layer. Lining thickness exhibits a dual effect: thicker linings generate fewer but wider cracks, whereas thinner linings produce more but narrower cracks. The proposed framework provides a reliable theoretical and engineering basis for safety assessment and design optimization of LRCs in CAES applications.</div></div>","PeriodicalId":49414,"journal":{"name":"Tunnelling and Underground Space Technology","volume":"172 ","pages":"Article 107460"},"PeriodicalIF":7.4,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146098893","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Impact of aftershocks on the response of a post-mainshock damaged metro station structure in seismic subsidence site","authors":"Zhong-Liang Zhang ,&nbsp;Zhen-Dong Cui ,&nbsp;Pengpeng He ,&nbsp;Ronald Y.S. Pak","doi":"10.1016/j.tust.2026.107456","DOIUrl":"10.1016/j.tust.2026.107456","url":null,"abstract":"<div><div>This study investigates the impact of aftershocks on the seismic response of a post-mainshock damaged metro station structure, with a particular focus on the complex dynamic characteristics of seismic subsidence sites. A three-dimensional finite element model was developed to replicate the collapse evolution of a post-mainshock damaged metro station under aftershocks. The results show that under strong mainshocks, the aftershock-induced displacement increment ratio can reach 1.37. Even following a weak mainshock, aftershocks can trigger approximately 40% additional site subsidence. The structural uplift increment ratio decreases with increasing aftershock intensity ratio, with an average value of about 4.4%. The EPWP increment ratio can reach up to 2.4 during aftershocks. Notably, the damage evolution of metro stations exhibits a mainshock threshold effect, i.e., stronger mainshocks lead to earlier damage initiation, with damage ratios exceeding 30%. Critically, aftershocks can exacerbate the damage, forming pervasive damage zones. Importantly, the inter-story drift shows a positive correlation with the damage ratio, surrounding soil displacement increment ratio, and EPWP increment ratio. A modified damage index is proposed to accurately evaluate structural damage under mainshock-aftershock sequences. The findings provide a valuable reference for the seismic design and post-earthquake rescue of metro stations in urban soft soil areas.</div></div>","PeriodicalId":49414,"journal":{"name":"Tunnelling and Underground Space Technology","volume":"171 ","pages":"Article 107456"},"PeriodicalIF":7.4,"publicationDate":"2026-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145962609","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A damage constitutive model of gas-bearing coal under pre-static loading and cyclic impact","authors":"Wei Wang ,&nbsp;Hanpeng Wang ,&nbsp;Xinyuan Xie ,&nbsp;Zicheng Wang ,&nbsp;Weibing Cai ,&nbsp;Yunhao Wu ,&nbsp;Yuguo Zhou","doi":"10.1016/j.tust.2025.107421","DOIUrl":"10.1016/j.tust.2025.107421","url":null,"abstract":"<div><div>The increasing depth of coal mining has led to more severe dynamic disasters, such as coal and gas outbursts, under complex environments characterized by high in-situ stress, elevated gas pressure, and cyclic excavation-induced disturbances. However, existing damage constitutive models rarely comprehensively consider the combined effects of static loading, gas, and cyclic impact. To address this, a novel time-dependent damage constitutive model for gas-bearing coal under pre-static loading and cyclic impact is developed in this study. Based on the framework of the generalized Kelvin model, the elastic elements are replaced with damage elements. Following the strain equivalence principle, a coupling damage factor integrating static loading and gas effects is derived. The cyclic impact-induced damage, considering strain rate effects, is represented by a parallel configuration of a damage element and a viscous element. Meanwhile, an inverted S-shaped cyclic impact damage factor is established based on the inverse logistic function, effectively capturing the three-stage damage evolution (initial rapid increase, stabilization, and subsequent acceleration) by incorporating the effects of impact number, frequency, and peak amplitude. Numerical simulations of a coal and gas outburst induced by roadway excavation with cyclic disturbance are conducted using the proposed model. The results demonstrate consistency with physical simulations under identical conditions regarding stress evolution, gas pressure variation, and outburst cavity location, confirming the validity and applicability of the proposed damage constitutive model. The proposed model can accurately capture the inherent laws of damage evolution of gas containing coal under complex loads, providing a theoretical tool for understanding and predicting dynamic disasters induced by cyclic disturbances in deep mining.</div></div>","PeriodicalId":49414,"journal":{"name":"Tunnelling and Underground Space Technology","volume":"171 ","pages":"Article 107421"},"PeriodicalIF":7.4,"publicationDate":"2026-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145928418","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Intelligent prediction of surface settlement troughs induced by twin shields tunnelling: Insights from a numerical modelling-empirical formulation-interpretable automated machine learning fusion method","authors":"Chengwen Wang,&nbsp;Xiaoli Liu,&nbsp;Weiqiang Xie,&nbsp;Yanlin Su,&nbsp;Yingtong Ju","doi":"10.1016/j.tust.2026.107449","DOIUrl":"10.1016/j.tust.2026.107449","url":null,"abstract":"<div><div>The construction of twin shield tunnels has become increasingly prevalent in densely populated urban areas. Accurately predicting the surface settlement induced by twin-shield tunnelling is of great significance for risk mitigation and refined settlement control. This study proposes a novel intelligent approach that integrates numerical modelling, empirical formula, and automated machine learning (AutoML) to predict surface settlement troughs induced by twin-shield tunnelling. Using a well-validated numerical model that considered 11 input parameters (including geological, geometric, and operational factors), 2000 settlement trough datasets were generated through numerical modelling. Subsequently, an improved superposition method was applied to extract six characteristic control parameters of the settlement troughs, thereby constructing a high-quality dataset. A multi-output AutoML model was then developed to predict the control parameters of the twin-tunnel-induced settlement troughs. Compared with six conventional machine learning models and two classical ensemble strategies, the AutoML model exhibited superior predictive accuracy and generalization capability, achieving average <em>R</em>² values of 0.9977 and 0.9835 for the training and test sets, respectively. The Shapley Additive Explanations (SHAP) method was employed to analyze the interpretability of the AutoML model. The results highlight the significant influence of construction parameters (e.g., tunnelling contraction ratio) on the maximum settlement, as well as the regulatory effects of geometric parameters (tunnel diameter, burial depth, and twin-tunnel spacing) on the shape of the settlement trough, thereby providing valuable guidance for design optimization and precise construction control. Finally, the proposed AutoML model was validated using five real-world engineering cases, where the predicted settlement troughs closely matched the measured data, thereby confirming the robustness, reliability, and practical applicability of the model and demonstrating its promising potential for engineering practice.</div></div>","PeriodicalId":49414,"journal":{"name":"Tunnelling and Underground Space Technology","volume":"171 ","pages":"Article 107449"},"PeriodicalIF":7.4,"publicationDate":"2026-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145928515","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"New sensing-inversion integrated method for mechanical behavior analysis of shield tunnels during heavy rainfall","authors":"Haoran Wang ,&nbsp;Chengchao Guo ,&nbsp;DingFeng Cao ,&nbsp;Jin Tang ,&nbsp;Fuming Wang","doi":"10.1016/j.tust.2025.107441","DOIUrl":"10.1016/j.tust.2025.107441","url":null,"abstract":"<div><div>In this study, a sensing-inversion method was proposed to investigate the mechanical response mechanisms of shield tunnels under heavy rainfall conditions, integrating displacement monitoring, distributed fiber optic sensing, and a strain–displacement-internal force recursive inversion method. Physical model tests were conducted to simulate interactions between heavy rainfall, soil strata, and tunnel structures. Laser displacement sensors and distributed optical fibers were used to monitor dynamic structural deformations and strains. An inversion model based on elastic foundation curved beam theory was developed to quantitatively analyze tunnel deformation evolution, load development mechanisms, and internal force distribution characteristics. The results indicate that the proposed inversion method improved accuracy by over 80% compared to conventional models and effectively captured radial displacements and internal force distributions. Under rainfall loading, the tunnel lining exhibited elliptical deformation and settlement, accompanied by compressive stresses at the crown and invert. The region of compressive stress expanded with increasing overburden thickness, whereas tensile stress developed at the haunches. The compressive stress at the crown exceeded that at the invert. When the tunnel was deeply buried, longer rainfall infiltration paths delayed structural responses to water penetration. Furthermore, deep overburden facilitated the dispersion localized stress concentrations in the lining caused by rainfall.</div></div>","PeriodicalId":49414,"journal":{"name":"Tunnelling and Underground Space Technology","volume":"171 ","pages":"Article 107441"},"PeriodicalIF":7.4,"publicationDate":"2026-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145928513","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Dynamic prediction of surrounding rock grades in TBM tunnels based on physics–data dual-driven model","authors":"Kang Fu ,&nbsp;Yiguo Xue ,&nbsp;Daohong Qiu ,&nbsp;Fanmeng Kong ,&nbsp;Jianning Wang","doi":"10.1016/j.tust.2025.107311","DOIUrl":"10.1016/j.tust.2025.107311","url":null,"abstract":"<div><div>Accurate and dynamic identification of surrounding rock grades in TBM tunnels is crucial for ensuring excavation safety and improving construction efficiency. This study proposes a hybrid modeling method based on a physics-data dual-driven approach to achieve high-precision dynamic identification of surrounding rock grades. First, the Isolation Forest model is employed to eliminate outliers from the raw tunneling data, and the key tunneling parameters influencing rock grades are identified using mutual information. Then, the Seasonal and Trend decomposition using LOESS (STL) model is used to perform multimodal decomposition on the dominant tunneling parameters, obtaining the corresponding trend, periodic, and residual components. Subsequently, an Improved Refined Composite Multiscale Sample Entropy (IRCMSE) model is adopted to calculate the feature entropy of each component, forming a dynamic sample database for the data-driven model. Based on this, an improved Convolutional Neural Network – Long Short-Term Memory (CNN-LSTM) model is developed to realize data-driven dynamic identification of TBM tunnel strata. Furthermore, a variation identification formula for surrounding rock grades was proposed based on the principle of geological continuity, enabling physics-driven dynamic identification of surrounding rock grades in TBM tunnels. On this basis, a fusion method combining the physical-driven model and the data-driven model is proposed. The constructed physics-data dual-driven model achieves average precision, recall, F1-score, and accuracy of 98.29 %, 97.98 %, 98.13 %, and 98.30 %, respectively, representing an average improvement of 2.17 % over the data-driven model and 15.07 % over the physical-driven model. Engineering validation results indicate that the overall performance of the model decreases by only 1.74 % and 5.29 % under similar and different geological conditions, respectively, demonstrating strong generalization and robustness, and meeting the requirements of intelligent TBM tunneling under complex geological conditions.</div></div>","PeriodicalId":49414,"journal":{"name":"Tunnelling and Underground Space Technology","volume":"171 ","pages":"Article 107311"},"PeriodicalIF":7.4,"publicationDate":"2026-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145897494","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Optimized CNN-BiLSTM-Attention with hybrid signal denoising: a novel interpretable framework for prediction of shield tunneling advance speed","authors":"Wei Jin ,&nbsp;Kangping Gao ,&nbsp;Chengyao Liu","doi":"10.1016/j.tust.2026.107471","DOIUrl":"10.1016/j.tust.2026.107471","url":null,"abstract":"<div><div>To address the challenges of insufficient prediction accuracy and poor stability of shield tunneling advance speed (AS), this study proposes an intelligent prediction framework based on deep learning. First, a comprehensive data preprocessing strategy is applied, integrating boxplot-based outlier removal, sliding-window smoothing, and a hybrid denoising method combining ensemble empirical mode decomposition (EEMD) with sample entropy-weighted wavelet thresholding. This strategy effectively corrects raw monitoring data, enhances stationarity and signal-to-noise ratio, with its efficacy confirmed through ablation experiments. Subsequently, four key input features are selected from multi-source TBM operational parameters using Pearson correlation analysis. Building upon this, a novel CNN-BiLSTM-Attention model is constructed by synergistically integrating convolutional neural networks (CNN), bidirectional long short-term memory (BiLSTM) networks, and an attention mechanism. This architecture facilitates the collaborative extraction of local spatial features and the modeling of long-term temporal dependencies. Furthermore, the Optuna framework is introduced for automated hyperparameter optimization to configure the model structure. Results demonstrate that the optimized model achieves significant performance improvements: the coefficient of determination (R²) and variance accounted for (VAF) increase from 0.84 to 0.94, while the root mean square error (RMSE), mean absolute error (MAE), and mean absolute percentage error (MAPE) decrease by 0.61, 0.49, and 1.42%, respectively. Compared to benchmark models such as GA-LightGBM, CNN-LSTM, and XGBoost, the proposed model demonstrates superior performance, with R² and VAF improving by at least 0.11, and RMSE, MAE, and MAPE decreasing by at least 0.63, 0.33, and 1.13%, respectively. The proposed model also slightly outperforms more advanced Transformer models. SHAP interpretability analysis confirms the validity of the feature selection and quantifies parameter contributions, identifying cutterhead penetration and torque as the most influential factors for advance speed prediction. Overall, the proposed model demonstrates stable and superior performance in terms of prediction accuracy and generalization capability.</div></div>","PeriodicalId":49414,"journal":{"name":"Tunnelling and Underground Space Technology","volume":"171 ","pages":"Article 107471"},"PeriodicalIF":7.4,"publicationDate":"2026-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146071730","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Short-term surface settlements induced by EPBM twin tunnelling in saturated sandy soils","authors":"Binh Thanh Le ,&nbsp;Sam Divall ,&nbsp;Tra Nguyen ,&nbsp;Michael C.R. Davies","doi":"10.1016/j.tust.2026.107482","DOIUrl":"10.1016/j.tust.2026.107482","url":null,"abstract":"<div><div>This paper presents a case study of the construction of a 781-metre-long twin-tunnel, using an Earth Pressure Balance Machine (EPBM), in saturated sandy soils in Ho Chi Minh City, Vietnam. The depths of the tunnels were between 11.4m and 24.6m below the ground surface. The averaged trough width and length parameters were 0.326 and 0.446, which are consistent with previous studies in sands. The volume losses ranged from the anticipated levels of less than 0.5% to notably high values reaching up to 2.44%. Low volume losses were associated with areas of dense soil and effective tail void grouting. The characteristics of effective tail void grouting observed in dense sand in this project were grouting pressures close to porewater pressure, coupled with stable grouting volume that was approximately 130% of the volume of the theoretical tail void for the majority of the drive. However, in very loose sandy soil zones it was observed that even very high tail void grout volume did not prevent large settlements. Soil relative density proved to be an influential factor in ground surface vertical displacement, with large magnitudes occurring mainly in loose soil. A threshold relative density <span><math><mrow><msub><mi>I</mi><mi>d</mi></msub><mo>≈</mo><mn>0.4</mn></mrow></math></span> divides the normal volume loss of less than 0.7% range, and that of considerably larger volume loss. The results emphasised the need for caution when tunnelling at shallow depth in loose soil, where the combination of low relative density and shallow cover can result in significant ground movements.</div></div>","PeriodicalId":49414,"journal":{"name":"Tunnelling and Underground Space Technology","volume":"171 ","pages":"Article 107482"},"PeriodicalIF":7.4,"publicationDate":"2026-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146071737","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Full-scale experimental study of flame behavior and thermal distribution in utility tunnel fires","authors":"Xiaojiang Liu ,&nbsp;Zhao-Dong Xu ,&nbsp;Jun Dai ,&nbsp;Jiayu Che ,&nbsp;Zhong-Wei Hu ,&nbsp;Defu Che","doi":"10.1016/j.tust.2025.107439","DOIUrl":"10.1016/j.tust.2025.107439","url":null,"abstract":"<div><div>A series of full-scale fire experiments were conducted in the largest utility tunnel platform in China (100 m × 3 m × 3 m), aiming to investigate the flame behavior and thermal flow evolution of utility tunnel fires. Both oil pool fires and cable fires were employed to realistically replicate fire scenes while maintaining reasonable costs. The heat release rates, ventilation conditions, multiple fire sources, and cable arrangements were analyzed comprehensively. Oil pool fires were conducted to simulate the thermal output of early-stage cable fires, results under ventilated conditions demonstrated downstream shifts in peak temperatures and asymmetric longitudinal temperature distributions due to interactions between hot smoke and opposing airflow. In cable fires, heat release accumulates through layer-by-layer ignition pattern. The flame initially propagates vertically, then transitions to longitudinal spread after sufficient thermal accumulation. Based on these findings, shutting down ventilation and sealing openings during early fire stages is recommended to limit oxygen supply, as analyzed from ventilation tests under both oil pool and cable fire cases. Optimizing cable layout can further mitigate vertical flame spread, such as placing cables in lower bracket layers and adding fire-resistant partitions, based on the results from cable fire tests. These results provide primary full-scale experimental data and new insights into the distinct fire dynamics and thermal behavior of spreading fire sources in elongated, confined utility tunnels, offering valuable references for fire safety design and risk control in similar infrastructures.</div></div>","PeriodicalId":49414,"journal":{"name":"Tunnelling and Underground Space Technology","volume":"171 ","pages":"Article 107439"},"PeriodicalIF":7.4,"publicationDate":"2026-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145886126","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Multiple-branching hierarchical fusion network for tunnel blasting excavation shape prediction using measurement while drilling data","authors":"Jun Wang ,&nbsp;Qian Fang ,&nbsp;Jinkun Huang ,&nbsp;Guoli Zheng","doi":"10.1016/j.tust.2026.107479","DOIUrl":"10.1016/j.tust.2026.107479","url":null,"abstract":"<div><div>Tunnel excavation shape characteristics using the drill-and-blast method are influenced by blasting parameters and geological conditions. Meanwhile, these excavation shape characteristics significantly impact tunnel stability, support construction quality, and construction costs. Traditional approaches for predicting excavation shape characteristics exhibit limitations in terms of accuracy and efficiency. To address this, we propose a multiple-branching hierarchical fusion extraction network (MBHF) for predicting tunnel excavation shape characteristics, including overbreak/underbreak values, cross-sectional curvature, and regional convexity/concavity. The MBHF model employs a multi-branch module for multi-attribute feature extraction and a gated structure for multi-source data fusion. It integrates construction parameters, blasting parameters, and measurement-while-drilling (MWD) data as inputs. The proposed MBHF model’s performance and superiority in feature extraction and fusion have been validated through ablation studies and comparative analyses. On the testing dataset, the model achieved high predictive accuracy, with R2 values of 0.97, 0.89, and 0.88 across the three prediction tasks. The multi-task learning strategy demonstrates superior performance compared to the single-task learning strategy in predicting excavation shape characteristics. The weight-constrained graph convolutional networks exhibit exceptional performance than the traditional graph convolutional networks in extracting features from MWD data. Increasing MWD data points (blasthole number and length) significantly enhances model performance, but performance declines beyond a certain threshold.</div></div>","PeriodicalId":49414,"journal":{"name":"Tunnelling and Underground Space Technology","volume":"171 ","pages":"Article 107479"},"PeriodicalIF":7.4,"publicationDate":"2026-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146033336","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}