Graph-convolutional neural networks for predicting tunnel boring machine performance

Abstract The development of human civilization is characterized by continuous breakthroughs in spatiotemporal constraints, and manufacturing has been central throughout the development. As a species capable of making and using tools, humanity continuously creates and transforms the material world. The essence of manufacturing is converting "resources" into "products". With the advancement of civilization, these resources have expanded from readily available stones and wood to various advanced materials and even individual atoms. Concurrently, the products have likewise evolved from simple hunting and gathering tools to tunnel boring machines, nanochips, intelligent robots, and transplantable organs. This pursuit of the "extreme" has given rise to a new manufacturing paradigm: Extreme Manufacturing (EM).

Introduction: The cutter is a crucial excavation tool employed by Tunnel Boring Machines (TBM) for tunneling underground passages. During operation, it undergoes intense impact loads, which are transmitted to the bearings, thus posing a risk of bearing failure. Method: This paper combines the Discrete Element Method (DEM) and Multibody Dynamics (MBD) by establishing a coupled simulation patent model of the cutter and surrounding rock, various methods of cutter load profiles are compared. The study investigates the variation patterns of multi-directional rock-breaking loads on the cutter and validates the findings through wire-cutting experiments. Result: The research findings indicate that the discrepancies between the simulated and experimental mean values of the cutter's normal force and rolling force are 6.26% and 35.20%, respectively. The efficiency of cutter load transmission to the outer ring of the bearing is 99.10%, leading to the characterization of the vibration characteristics curve of the bearing's outer ring. Simultaneously, the mean square error of the cutter load obtained through the coupled method and traditional method is 78.50 kN and 76.10 kN, respectively. In comparison, the experimental load's mean square error is 99.20 kN, indicating that the coupled method better aligns with actual conditions. Conclusion: This research approach serves as a reference for TBM cutter performance analysis and bearing fatigue analysis.

Abstract Excavation with tunnel boring machines (TBMs) has been extensively used in the construction of tunnelling projects. The geological conditions of the Himalayan region pose significant challenges to the performance of TBM tunnelling. This manuscript proposes a machine learning (ML)-based data-driven approach to predict the TBM net penetration rate (PRnet), integrating both geological and TBM operational parameters. To do so, a total of 8,614 stable-phase real-time TBM cycle data corresponding to mapped geological parameters were collected from a 12 km long tunnel in Nepal. The preprocessed dataset were randomly split into a training and testing set with the 80/20 rule. The TBM PRnet is evaluated using different non-ensemble, ensemble, and artificial neural network (ANN) regression models. The ANN and stacking ensemble models achieved the highest R 2 value of 0.94 on unseen test data. Shapley Additive exPlanations (SHAP) method, an explainable artificial intelligence tool, was used to analyze the influence of input features. The output results have revealed that PRnet is significantly influenced by both geological and TBM operational parameters. These parameters were further used to examine TBM jamming in areas where geological challenges associated to faults or weak zones were encountered. The evaluation results revealed that fluctuations in torque and thrust provide valuable information for assessing the risk of operational hazards in challenging geological environments. The manuscript offers a more adaptable prediction framework. Thus, the authors emphasize that the effective application of the ML-based data-driven approach in TBM tunnelling holds substantial potential for accurately predicting the net penetration rate. Highlights Machine learning (ML)-based model developed to predict TBM net penetration rate (PRnet). TBM operational parameters and geological conditions are utilized as input parameters. Prediction model incorporates data anomalies associated with complex geological environment. Impact of input features on TBM net penetration rate evaluated using SHAP method. Risk of TBM jamming assessed based on machine parameters and geological conditions.

In the context of the active development of the metro system, special attention is given to ensuring the safety of existing buildings and structures located in the construction impact zone. The construction of new tunnels near operational metro facilities may lead to undesirable deformations, including the subsidence of structures or rail threads. The aim of this study is to identify the key factors influencing the subsidence of the shallow station structure of the Moscow Metro during the construction of a tunnel using a tunneling boring machine (TBM) directly beneath it. The initial data are based on results from geotechnical monitoring, compared with the technological parameters of tunneling – the average ground support pressure at the face (p), the volume of ground output for installing one ring (vg), the volume of injected grout into annular void (vi) – as well as geometric characteristics (plan distance (r) and height (h) to the observation point) and the physical-mechanical properties of the soils. It was found that the most significant influence on the subsidence magnitude comes from the following parameters (in descending order of importance): r, vg, p, vi. Based on the identified factors, a subsidence forecasting model was constructed, explaining 90,9% of the sample variance and having a mean squared error (MSE) of 0,1353 mm2, which confirms its high predictive accuracy and its adequacy for practical application.

Rapid, accurate, and efficient prediction of surrounding rock grades is crucial for ensuring the safety and enhancing the efficiency of tunnel boring machine (TBM) construction. To achieve intelligent perception of surrounding rock grades based on TBM tunneling parameters, this study leverages data from the TBM1 construction phase of the Luotian Reservoir-Tiegang Reservoir Water Diversion Tunnel Project, integrating geological records and tunneling parameters to establish models for different rock grades. First, raw data were cleaned and denoised using box plots, followed by the selection of eight critical parameters—including thrust, torque, penetration rate (PR), rotation speed (RS), et al—through a hybrid approach combining “knowledge-driven” and “data-driven” criteria. The dataset was partitioned into training, testing, and validation sets at a 7:2:1 ratio. Three data processing methods were applied, and machine learning algorithms (XGBoost, Random Forest, CatBoost, and LightGBM) were employed to construct surrounding rock classification models, with Optuna hyperparameter optimization implemented to enhance model performance. The result reveals that the CatBoost model, optimized via SMOTE (Synthetic Minority Oversampling Technique) and hyperparameter tuning, delivered superior performance, achieving 99% validation accuracy with no misclassification across adjacent surrounding rock grades. This research provides actionable insights for advancing intelligent TBM construction practices.

This study presents a comprehensive jamming risk assessment framework for Tunnel Boring Machine (TBM) jamming accidents during excavation. Using real-time boring data and Bayesian conditional probability, a novel risk warning model is proposed to enhance safety and efficiency of tunneling projects. Through statistical analysis of excavation parameters, distinct patterns between jamming and normal excavation states are identified. A comprehensive jamming perception index (η) is introduced that synthesizes multiple parameters to accurately identify jamming states with a recognition rate of 95%. This integrated approach overcomes the limitations of single-parameter analysis and provides improved accuracy in jamming risk assessment. Additionally, a quantitative model for calculating jamming probability is developed, accounting for differences in sample size between jamming and normal excavation sections. The refined model yields realistic estimates of jamming probability, with an average of 94% in jamming sections and 7% in normal excavation sections. Furthermore, geological analysis shows that the Class Ⅲ surrounding rock is the most suitable for excavation and has the lowest jamming probability. This finding emphasizes the importance of considering geological conditions in excavation planning to effectively mitigate jamming risks. In conclusion, this research provides a practical framework for the prediction and management of TBM jamming accidents, contributing to enhanced safety and efficiency in tunneling projects.

Abstract Compressed air energy storage (CAES) has emerged as a grid‐scale energy storage linchpin, providing diurnal‐to‐seasonal timescale energy buffering for renewable power integration. Diverging from conventional salt cavern‐dependent approaches, artificial cavern‐based CAES unlocks geographical adaptability through engineered underground containment. This study systematically reviews critical technologies in chamber construction, including site selection, structural design, excavation methods, and post‐construction evaluation. Site selection employs a multi‐criteria matrix that combines geological and environmental factors. Structural design integrates spatial layout, burial depth, sealing system, and component compatibility to ensure chamber stability. Excavation prioritizes controlled blasting for homogeneous rock, while a tunnel boring machine is deployed in fractured zones to preserve integrity. Post‐construction assessments validate load‐bearing capacity, sealing performance, and operational readiness, supported by data‐driven maintenance strategies. Ongoing challenges include site‐specific geological risks, sealing system durability under cyclic loading, equipment integration, field‐scale validation, standardization gaps, and cost‐efficiency optimization. These innovations will establish best practices for building large‐scale, high‐efficiency CAES plants with ultra‐long duration and grid resilience, accelerating the transition to carbon‐neutral power systems.

Abstract This study introduces a novel integrated framework for real‐time tunnel boring machine (TBM) thrust prediction, addressing critical limitations in handling non‐stationarity, complex spatiotemporal dependencies, and abrupt disturbances. First, a real‐time windowed multi‐resolution analysis process, which performs decomposition strictly within each segmented sample window, is presented to explicitly disentangle the latent multi‐scale dependencies embedded in the thrust data. This ensures strict causality (using only current/historical data), prevents information leakage, and enhances resolution adaptability by capturing local dynamics specific to each data segment, overcoming global averaging effects. Second, a novel synergistic prediction architecture, integrating a hybrid static model with dynamic online residual correction, is proposed. A specifically optimized CNN‐LSTM‐attention primary model learns complex long‐term global patterns. Crucially, an efficient random Fourier features‐based online module is dedicated solely to real‐time learning of the primary model's residual dynamics, acting as a dynamic corrector rather than an independent predictor. This targeted residual correction significantly enhances robustness against non‐stationarity and disturbances. These innovations form an integrated solution and systematically address real‐time capability, local adaptability, complex pattern learning, and dynamic error correction. The results indicate that the presented method reduces the mean absolute percentage error from 2.84% to 1.89% and increased from 0.901 to 0.953. The generalizability of the model was further confirmed through the application of diverse datasets obtained from various chainages along the route. The proposed machine learning–based model can provide guidance for operators in real‐time TBM parameter adjustment during construction

Abstract Effective management of tunnel boring machine (TBM) jamming is crucial for ensuring safety and mitigating construction downtime. However, previous studies have primarily focused on predictive modeling based on numerical datasets, with limited consideration of field‐based geological conditions and inadequate investigation of the fundamental mechanisms underlying jamming phenomena. This study utilized two ensemble learning algorithms, Random Forest and Extreme Gradient Boosting, to predict TBM jamming based on a field dataset from 39 tunneling projects. A data augmentation technique was employed to construct an expanded dataset. The predictive model trained on the augmented dataset demonstrated improved detection of TBM jamming compared to the model developed without data augmentation. The jamming mechanism was successfully characterized, revealing the individual effects of geological factors and their complex interactions. A distinct difference in predictive uncertainty between correct and incorrect predictions supports the model's reliability. Finally, a practical risk management system was proposed by incorporating the predictive model with probability thresholds and validated through field application.

When tunnel boring machines (TBMs) excavate in mudstone, argillization of the rock reduces tunneling efficiency. To investigate the influence of argillization on TBM performance and determine the optimal operational parameters, numerical investigation was conducted based on the energy evolution by Particle Flow Code 3D (PFC3D). Argillization was simulated by adhesive rolling resistance Linear model. The effects of argillization on forces acting on the disc cutter, crack evolution, and energy consumption were analyzed. Furthermore, the influence of operational modes, tip angles, and tip widths on energy consumption, the mass of slurry adhered to the cutters, and tunneling efficiency were investigated. Results indicate that argillization decreases the normal force while increasing the rolling and lateral forces. Besides, argillization significantly increases mechanical work, thereby reducing tunneling efficiency of the TBM. When excavating in the mudstone, load control mode, coupled with a cutter tip angle of 40° and a tip width of 15mm, can effectively mitigate argillization risks and improve efficiency. This study provides valuable references for the operation of the TBM in mudstone, thereby expanding the machine's range of application.

Prediction of rock-breaking forces of tunnel boring machine (TBM) disc cutter based on machine learning methods

Wear anomaly detection method of tunnel boring machine disc cutters based on anomaly-attention improved long short-term memory autoencoder

Steel arches are an important support structure for tunnel boring machines (TBM) to pass through the broken formation, each piece of which has a mass of more than 200 kg. Traditional manual looping of steel arch is difficult, which can easily lead to untimely support or unstable quality, and even cause collapse accidents. It is necessary to design a simple and reliable steel arch looping mechanism. In this paper, the algebraic solution model of single-degree-of-freedom planar closed-chain mechanism is firstly established, and the design method of closed-chain end-effector based on input constraints and output requirements is proposed, based on which, twelve tree-type opened and closed chain steel arch looping mechanism configurations are deduced. Then, in the absence of dimension parameters, a performance evaluation model of the open and closed chain mechanism containing motion/force transfer performance and complexity dimensions is established, and a method for optimizing the steel arch looping mechanism configurations is proposed from the perspective of topology. Finally, the prototype experiment of steel arch looping operation was carried out to verify the feasibility of the optimal configuration of steel arch looping mechanism, which can provide a theoretical basis for the configuration design of other tree-type open and closed chain mechanisms.

TBM big data preprocessing method in machine learning and its application to tunneling

To enhance the measurement accuracy of Tunnel Boring Machine (TBM) guidance systems under strong vibration environments, this study designed a novel multi-degree-of-freedom (MDOF) active vibration isolation system. The system employs a composite structure of metal coil springs and eight large-air-gap voice coil actuators. This design effectively addresses the insufficient suppression of MDOF, large-amplitude (2 mm), low-frequency (5 Hz) strong vibrations by existing passive isolation systems and active micro-vibration isolation systems. Using modal analysis, the strong coupling effects between multiple channels were eliminated (coupling bandwidth reduced from >5.81 Hz to 0 Hz). Based on this decoupling, an Extended State Observer (ESO) and a high-frequency robust control algorithm were designed, significantly improving the system’s ability to suppress low-frequency disturbances. Tests show that under sweep excitation in the X/Y/Z directions, the system reduced the RMS measurement errors of pitch angle and roll angle by 78% and 74.8% respectively, and reduced the RMS linear vibrations in the X/Y/Z directions by 89.37%, 80.24%, and 94.92% respectively. The novel MDOF active vibration isolation system designed in this study provides a foundation for stable and high-precision operation of TBM guidance systems in complex geological vibration environments.

Uncertainty design optimization of the main bearing in tunnel boring machine based on the Kriging model with partial least squares

Decision tree analysis of cutter selection for tunnel boring machines: A study of geological conditions and machine types in high-performing TBM projects

The main bearing of a tunnel boring machine (TBM) is a critical component of the main driving system that enables continuous excavation, and its performance is crucial for ensuring the safe operation of the TBM. Currently, there are few testing technologies for TBM main bearings, and a comprehensive testing and evaluation system has yet to be established. This study presents an experimental investigation using a self-developed, full-scale TBM main bearing test bench. Based on a representative load spectrum, both operational condition tests and life cycle tests are conducted alternately, during which the signals of the main bearing are collected. The observed vibration signals are weak, with significant vibration attenuation occurring in the large structural components. Compared with the test bearing, which reaches a vibration amplitude of 10 g in scale tests, the difference is several orders of magnitude smaller. To effectively utilize the selected evaluation indicators, the entropy weight method is employed to assign weights to the indicators, and a comprehensive analysis is conducted using grey relational analysis. This strategy results in the development of a comprehensive evaluation method based on entropy weighting and grey relational analysis. The main bearing performance is evaluated under various working conditions and the same working conditions in different time periods. The results show that the greater the bearing load, the lower the comprehensive evaluation coefficient of bearing performance. A multistage evaluation method is adopted to evaluate the performance and condition of the main bearing across multiple working scenarios. With the increase of the test duration, the bearing performance exhibits gradual degradation, aligning with the expected outcomes. The findings demonstrate that the proposed performance evaluation method can effectively and accurately evaluate the performance of TBM main bearings, providing theoretical and technical support for the safe operation of TBMs.

Introduction Tunnel construction is a high-risk, complex task requiring precision, safety, and efficiency. With growing infrastructure demands, this study proposes a hybrid framework integrating Building Information Modeling (BIM), machine learning models such as Artificial Neural Network (ANN), K-Nearest neighbors (KNN), Support Vector Machines (SVM), and advanced optimization techniques to improve decision-making, predict geological challenges, and automate key operations in large-diameter tunnel projects, enhancing overall project performance and risk management. Methods Various methods are employed in the study, including BIM, machine learning, and robust optimization, which can be perceived as enhancing tunnel construction. Prediction using AI-based algorithms, namely ANN, KNN, and SVM, was made possible with real-time sensor data on geological issues. FANUC ROBOGUIDE software was also used to simulate the actions of robots, ensuring that material handling was performed with precision. Among these three, the optimal performance of SVM outshines ANN and KNN. Results The results have shown that BIM integrated with machine learning and optimization significantly increased tunnel construction performance. In predicting critical operational parameters, AI-based models, especially SVM, were found to provide an accuracy of 98.56%, outperforming KNN and ANN. Hence, this kind of predictability may allow for real-time modifications in the Tunnel Boring Machines (TBM) settings, thereby decreasing the risks associated with geological uncertainties. Additionally, the FANUC ROBOGUIDE software will ensure more precise and collision-free material handling, further enhancing safety and efficiency in tunnel construction projects. Discussion The study demonstrates that integrating BIM with machine learning and robotic simulation significantly enhances tunnel construction efficiency and safety. Among the models evaluated, SVM achieved the highest accuracy (98.56%) in predicting geological challenges. Real-time data processing enabled timely adjustments to TBM operations, while FANUC ROBOGUIDE ensured precise material handling, reducing risks and delays in complex construction environments. Conclusion The research currently underway has established the efficacy of integrating BIM, machine learning, and optimization in improving tunnel construction. The applications of AI models, such as SVM, KNN, and ANN, have improved targeted operational parameters and reduced geological risks, with SVM yielding the highest accuracy at 98.56%. Efficiency and safety were further enhanced by real-time data-driven decisions and robotic simulations. The developed framework offers a practical solution for enhancing decision-making and operational efficiency in complex engineering projects.