Data-Driven Approach for Intelligent Classification of Tunnel Surrounding Rock Using Integrated Fractal and Machine Learning Methods DOI Creative Commons
Junjie Ma, Tianbin Li, Roohollah Shirani Faradonbeh

et al.

Fractal and Fractional, Journal Year: 2024, Volume and Issue: 8(12), P. 677 - 677

Published: Nov. 21, 2024

The degree of rock mass discontinuity is crucial for evaluating surrounding quality, yet its accurate and rapid measurement at construction sites remains challenging. This study utilizes fractal dimension to characterize the geometric characteristics develops a data-driven classification (SRC) model integrating machine learning algorithms. Initially, box-counting method was introduced calculate from excavation face image. Subsequently, parameters affecting quality were analyzed selected, including strength, discontinuity, condition, in-situ stress groundwater orientation. compiled database containing 246 railway highway tunnel cases based on these parameters. Then, four SRC models constructed, Bayesian optimization (BO) with support vector (SVM), random forest (RF), adaptive boosting (AdaBoost), gradient decision tree (GBDT) Evaluation indicators, 5-fold cross-validation, precision, recall, F1-score, micro-F1-score, macro-F1-score, accuracy, receiver operating characteristic curve, demonstrated GBDT-BO model’s superior robustness in generalization compared other models. Furthermore, additional validated intelligent approach’s practicality. Finally, synthetic minority over-sampling technique employed balance training set. Subsequent retraining evaluation confirmed that imbalanced dataset does not adversely affect performance. proposed shows promise predicting guiding dynamic support.

Language: Английский

Bidirectional denoising method based on Fast Fourier transform analysis for TBM field penetration data DOI

Wenkun Yang,

Zuyu Chen, Haitao Zhao

et al.

Tunnelling and Underground Space Technology, Journal Year: 2025, Volume and Issue: 158, P. 106436 - 106436

Published: Feb. 6, 2025

Language: Английский

Citations

1
Intelligent Robust Control of Roadheader Based on Disturbance Observer DOI Creative Commons

Shuo Wang,

Dongjie Wang,

Aixiang Ma

et al.

Actuators, Journal Year: 2025, Volume and Issue: 14(1), P. 36 - 36

Published: Jan. 17, 2025

The formation of a coal mine roadway cross-section is primary task the boom-type roadheader. This paper proposes an intelligent robust control scheme for cutting head trajectory tunneling robot, which susceptible to unknown external disturbances, system nonlinearity, and parameter uncertainties. First, working conditions section were analyzed, mathematical model was established. Then, high-gain disturbance observer designed based on analyze loads compensate uncertainties disturbances. A sliding mode controller proposed using backstepping design method, incorporating saturation function term avoid chattering. eel foraging optimization algorithm also improved used tune parameters. simulation developed performance comparison tests. Finally, experimental verification conducted under actual in tunnel face, results demonstrated effectiveness method.

Language: Английский

Citations

0
Role of hole depth on mechanical behavior and acoustic emission characteristics of pre-drilled sandstone DOI Creative Commons

Jiahan Liu,

Ruide Lei

Frontiers in Earth Science, Journal Year: 2025, Volume and Issue: 13

Published: Jan. 23, 2025

To examine the influence of hole depth on mechanical properties rock, a series uniaxial compression tests were performed six groups pre-drilled sandstone samples, each with varying depths. Also, multiple physical fields coupled acoustic emission (AE) and digital image correlation (DIC) systems synchronously employed to monitor fracturing process. The study focused characterizing cracking fracturing, energy evolution, fracture patterns in sandstones different findings show that peak strength decreases linearly increase depth. mode transits from simple unilateral spalling complex characterized by fractures spalling. AE analysis shows deeper borehole, lower signal frequency, indicating fewer but more significant events. With depth, elastic sample 29.81 kJ/m 3 22.65 , dissipated increases 4.48 6.25 . Moreover, displays distinct multifractal spectrum features under stress levels. width (Δ α ) varies 0.419 0.227, suggesting small-scale events predominantly govern failure mechanism. DIC observation major principal strain concentration mainly occurs around hole. monitoring points cumulative at P2 P6 is significantly higher compared other regions. Furthermore, it observed release pathways originating newly formed cracks dislocation slips become diversified,

Language: Английский

Citations

0
Application of multifractal spectrum to characterize the evolution of multiple cracks in concrete beams DOI
Chunxiao Li, Maosen Cao, Xinxin Zhao

et al.

Structures, Journal Year: 2024, Volume and Issue: 70, P. 107868 - 107868

Published: Nov. 26, 2024

Language: Английский

Citations

1
Data-Driven Approach for Intelligent Classification of Tunnel Surrounding Rock Using Integrated Fractal and Machine Learning Methods DOI Creative Commons
Junjie Ma, Tianbin Li, Roohollah Shirani Faradonbeh

et al.

Fractal and Fractional, Journal Year: 2024, Volume and Issue: 8(12), P. 677 - 677

Published: Nov. 21, 2024

The degree of rock mass discontinuity is crucial for evaluating surrounding quality, yet its accurate and rapid measurement at construction sites remains challenging. This study utilizes fractal dimension to characterize the geometric characteristics develops a data-driven classification (SRC) model integrating machine learning algorithms. Initially, box-counting method was introduced calculate from excavation face image. Subsequently, parameters affecting quality were analyzed selected, including strength, discontinuity, condition, in-situ stress groundwater orientation. compiled database containing 246 railway highway tunnel cases based on these parameters. Then, four SRC models constructed, Bayesian optimization (BO) with support vector (SVM), random forest (RF), adaptive boosting (AdaBoost), gradient decision tree (GBDT) Evaluation indicators, 5-fold cross-validation, precision, recall, F1-score, micro-F1-score, macro-F1-score, accuracy, receiver operating characteristic curve, demonstrated GBDT-BO model’s superior robustness in generalization compared other models. Furthermore, additional validated intelligent approach’s practicality. Finally, synthetic minority over-sampling technique employed balance training set. Subsequent retraining evaluation confirmed that imbalanced dataset does not adversely affect performance. proposed shows promise predicting guiding dynamic support.

Language: Английский

Citations

0