Surface Subsidence Prediction Research Articles

Explainable deep learning-based dynamic prediction of surface settlement considering temporal characteristics during deep excavation

Accurate prediction of surface settlement (SS) induced by deep foundation pit (DFP) excavation is challenging considering the complicated soil distributions and uncertain construction situations. Therefore, a Recurrent Gated Unit (GRU) neural network model incorporated with Variation Mode Decomposition (VMD) was developed for SS prediction. By data denoising with VMD and hyperparameter tuning using Bayesian Optimization (BO), the proposed BO-GRU model achieved dynamic and accurate prediction of SS caused by DFP excavation. Shapley Additive exPlanations (SHAP) analysis was then performed to enhance the interpretability of the BO-GRU model. The presented BO-GRU model was validated with a case study of Wuhan Metro Line 12 Shiqiao Station. The results indicate that: (1) The BO-GRU models with VMD data denoising technique achieved accurate prediction with the average RMSE of 0.0244, MAE of 0.0189, MAPE of 0.0017, R2 of 0.9894, and VAF of 99.1206. (2) The BO-GRU model outperformed the other five state-of-the-art models in predictive performance with robustness improvement of 13.5%, 12.0%, 50.0%, 29.2%, and 27.3%, separately, compared to the other five models. (3) The historical values of the SS make the most significant contributions to the outcomes of the BO-GRU models. In short, this study enhances the accurate and dynamic prediction of SS caused by DFP excavation, contributing to the safe execution of DFP projects.

Machine Learning–Finite Element Mesh Optimization-Based Modeling and Prediction of Excavation-Induced Shield Tunnel Ground Settlement

Ground settlement prediction for shield construction is highly important and challenging. This study introduces a machine learning algorithm combined with finite element numerical simulation, i.e., machine learning–finite element mesh optimization. For surface subsidence prediction, 16 combination models of ANN, KNN, RF and SVR were optimized by PSO, GA, BT and BO, involving raw data preprocessing, principal component analysis, hyperparameter selection and prediction accuracy evaluation. A subway shield tunneling project was analyzed, in which the meshes of finite element numerical models were discretized into different sizes from 1.0[Formula: see text]m to 2.0[Formula: see text]m. In total, 360 sets of data points were extracted from the simulation results, including stress, strain, shield jacking force, internal friction angle, cohesion force, and settlement, of which 252 data points were used as the input parameters of machine learning model. Analysis of average error rate of finite element–machine learning coupling models showed that the finite element model had the highest accuracy of settlement prediction when the mesh size of the finite element model was 1.4[Formula: see text]m, and the GA-SVR model had the highest accuracy and generalization ability in ground settlement prediction. This study highlights the uniqueness of machine learning–finite element mesh optimization model in application.

A prediction model for surface settlement during the construction of variable cross-section tunnels under existing structures based on stochastic medium theory

When predicting ground surface settlement caused by constructing a new tunnel beneath existing structures, traditional stochastic medium theory inadequately accounts for the effects of abrupt changes in the cross-sectional areas of variable-section tunnels and the presence of existing structures, potentially leading to significant errors. In this paper, an enhanced ground surface settlement prediction model, based on stochastic medium theory, is proposed. The model equates the settlement of the existing structure’s base slab to that of the overlying soil and divides the horseshoe-shaped tunnel cross-section into eight arc segments, which are calculated using a polar coordinate system. Furthermore, the model treats soil loss at the junctions of variable-section tunnels as a linear transition, introduces the concept of a linear transition segment for variable sections, and accounts for the superimposed effects of closely spaced twin-tunnel excavations. Based on this model, a general-purpose calculation program has been developed that enables the rapid prediction of ground surface deformation caused by a variable-section tunnel passing beneath existing structures, simply by inputting engineering parameters. Finally, the accuracy of the prediction model was validated through comparisons with field measurement data, finite element analysis results, and calculations based on traditional stochastic medium theory. The results indicate that the proposed prediction model demonstrates high consistency with field data and finite element analysis results, whereas traditional stochastic medium theory results exhibit significant errors. This model is scientifically valid and provides a reliable reference for predicting ground surface settlement in comparable variable-section metro tunnel construction projects.

Research on the application of a hybrid intelligent algorithm in reducing the negative response of shield tunneling construction

Traditional shield project control relies on manual monitoring and management, which suffers from poor reliability, lag, and other defects. In this paper, a hybrid algorithm based on Bayesian optimization (BO), light gradient enhancer (LGBM), and nondominated sorting genetic algorithm-III (NSGA-III) is proposed to realize intelligent control of shield construction safety. The BO-LGBM-NSGA-III is a multiobjective optimization algorithm that takes cutterhead wear and shield energy consumption as objective functions and the nonlinear mapping function of surface settlement and construction parameters as fitness functions. Based on the Pareto distribution of the optimized backing structure construction parameters, reasonable limits of the control parameters of the shield tunneling machine are obtained. The conclusions are as follows. (1) The R2 of the test set for surface subsidence prediction is 0.916, and the RMSE is 2.655. (2) The proposed method can achieve multiobjective optimization, and the surface settlement, cutter head wear, and shield energy consumption are optimized by 49.92%, 24.38%, and 41.88%, respectively. (3) By using this method, both prediction and optimization can be realized, and the optimal control scheme can be obtained. The negative impact of shield tunneling construction has been significantly reduced.

A new model for predicting surface subsidence of twin salt cavern gas storages with different shapes

A new model for predicting surface subsidence of twin salt cavern gas storages with different shapes

Research on the surface subsidence characteristics and prediction models caused by coal mining under the reverse fault

Predicting and understanding the phenomenon of surface subsidence caused by coal mining in working faces with faults are important issues for safe coal mining and efficient production. In numerical simulation experiments, it was found that the phenomenon of surface subsidence manifests when faults exist, and the degree of influence of faults with different dip angles on surface subsidence varies. This phenomenon is attributed to fault activation. According to the experimental results, the impact of faults with different dip angles on surface subsidence falls into three levels: level I for 35° faults, level II for 45° and 55° faults, and level III for 65° and 75° faults. Similarly, the relationship between the difficulty of fault activation and the dip angle of faults can be categorized as 35° faults prone to activation, 45° and 55° faults difficult to activate, and 65° and 75° faults not prone to activation. The probability integral correction model for fault mining, which integrates the surface subsidence values caused by fault-induced attenuation and the subsidence arising from separation spaces, was introduced, thereby constructing a surface subsidence prediction model. This proposed prediction model can accurately predict surface subsidence, with a root mean square error of 10.74 mm between the predicted and measured values, as validated using DInSAR results from the III 6301 working face in the Jincheng mining area.

Procedure Design and Reliability Analysis for Prediction of Surface Subsidence of a Metal Mine Induced by Block Caving Method—A Case Study of Pulang Copper Mine in China

Surface subsidence resulting from block caving mining causes considerable environmental and economic harm in mining areas, highlighting the critical need for accurate predictions of surface subsidence. Given the unique features of the block caving technique and the resemblance between the released ore pillars and the mining processes, this paper developed a lightweight model to forecast surface settlement utilizing the probability integration approach to address the issue of predicting surface settlement in metallic mines. This study focuses on the Pulang Copper Mine, situated in the northeast of Shangri-La County within the Yunnan Province, as a case example. This mine employs the block caving method, which results in substantial surface subsidence. A visual mining simulation program is designed to combine the ore mining plan with the prediction model, manage the ore output of each mining point in batches, treat the ore pillars released in the planning cycle as strip work, and simulate and calculate the surface area above the ore pillars settlement value. The calculated values of surface subsidence induced by ore drawing are then interpreted as the downward displacement of the surface subsidence beneath the strip workings. Furthermore, to verify the reliability of the model, three-dimensional laser point cloud data of the Pulang Copper Mine in recent years were collected, and the differences between the predicted surface and the measured surface were calculated and analyzed.

A physics-constrained neural network for predicting excavation-induced ground surface settlement in clay

Accurate prediction of ground surface settlement (GSS) adjacent to an excavation is important to prevent potential damage to the surrounding environment. Previous studies have extensively delved into this topic but all under the limitations of either imprecise theories or insufficient data. In the present study, we proposed a physics-constrained neural network (PhyNN) for predicting excavation-induced GSS to fully integrate the theory of elasticity with observations and make full use of the strong fitting ability of neural networks (NNs). This model incorporates an analytical solution as an additional regularization term in the loss function to guide the training of NN. Moreover, we introduced three trainable parameters into the analytical solution so that it can be adaptively modified during the training process. The performance of the proposed PhyNN model is verified using data from a case study project. Results show that our PhyNN model achieves higher prediction accuracy, better generalization ability, and robustness than the purely data-driven NN model when confronted with data containing noise and outliers. Remarkably, by incorporating physical constraints, the admissible solution space of PhyNN is significantly narrowed, leading to a substantial reduction in the need for the amount of training data. The proposed PhyNN can be utilized as a general framework for integrating physical constraints into data-driven machine-learning models.

Surface settlement induced by frictional force of epb shield tunneling in sandy gravels.

The excavation of Earth Pressure Balance (EPB) shield can be divided into two distinct stages, i.e. advancing and lining installation. The frictional force applied on surrounding soils reverses at these two stages, which is harmful to the settlement control. Based on Mindlin's method, a new model of surface settlement is derived to involve the reversed friction. A closed form formula is then obtained for the major type of metro tunnels. Main operational parameters are also used as input of the formula. Numerous operational data and measured settlements are collected from EPB tunnels of Chengdu Metro, Line 7. The proposed formula is validated against these field data in sandy gravels. It is shown that the new formula gives reasonable prediction of surface settlement along the tunnel sections. The accuracy of new formula is significantly higher than that of Peck's formula. This study provides a new vision in settlement control of EPB shield tunneling. The increase of chamber pressure will induce higher negative friction during the lining installation. Therefore, surface settlement of EPB tunneling cannot be controlled by just increasing chamber pressure. A balanced relationship between the chamber pressure and the thrust should be maintained instead.

Research on prediction method of coal mining surface subsidence based on MMF optimization model

Coal seam mining causes fracture and movement of overlying strata in goaf, and endangers the safety of surface structures and underground pipelines. Based on the engineering geological conditions of 22,122 working face in Cuncaota No.2 Coal Mine of China Shenhua Shendong Coal Group Co., Ltd. a similar material model test of mining overburden rock was carried out. The subsidence of overburden rock was obtained through the full-section strain data of distributed optical fiber technology, and the characteristics of mining surface subsidence were studied. The Weibull model was used to adjust the mathematical form of the first half of the surface subsidence curve via the MMF function. On this basis, the prediction model of coal seam mining surface subsidence was established, and the parameters of the prediction model of surface subsidence were determined. The test results show that with the advancement of coal seam mining, the fit goodness of the surface subsidence prediction curve based on the MMF optimization model reaches 0.987. Compared with the measured values, the relative error of the surface subsidence prediction model is reduced to less than 10%. The model displays good prediction accuracy. The time required for settlement stability in the prediction model is positively correlated with parameter a and negatively correlated with parameter b. The research results can be further extended to the prediction of overburden “three zones” subsidence, and provide a scientific basis for the evaluation of surface subsidence compression potential in coal mine goaf.

A profile function to predict surface subsidence for down-cut slicing panels into inclined coal seam of the Barapukuria coal mine, Bangladesh

Mining-induced subsidence, a typical consequence of underground mining, can cause environmental and structural problems in the mine area. Predicting surface subsidence for inclined and deep seams has been less emphasized than horizontal seams, mainly due to the complexities of extracting such coal seams. The prediction of surface subsidence and the dynamics of subsidence in down-cut slicing panels is more complex and crucial in practice. This study presents a novel profile function method for forecasting surface subsidence resulting from downward slicing panels in the inclined coal seam mining of the Barapukuria coal mine. The calculated maximum subsidence depth is 10.23 meters whereas the measured maximum water depth of the subsided lake of the Barapukuria coal mine is 10.5 meters. The results indicate a strong correlation between the predicted and the measured values.

Prediction of Surface Subsidence in Mining Areas Based on Ascending-Descending Orbits Small Baseline Subset InSAR and Neural Network Optimization Models.

Surface subsidence hazards in mining areas are common geological disasters involving issues such as vegetation degradation and ground collapse during the mining process, which also raise safety concerns. To address the accuracy issues of traditional prediction models and study methods for predicting subsidence in open-pit mining areas, this study first employed 91 scenes of Sentinel-1A ascending and descending orbits images to monitor long-term deformations of a phosphate mine in Anning City, Yunnan Province, southwestern China. It obtained annual average subsidence rates and cumulative surface deformation values for the study area. Subsequently, a two-dimensional deformation decomposition was conducted using a time-series registration interpolation method to determine the distribution of vertical and east-west deformations. Finally, three prediction models were employed: Back Propagation Neural Network (BPNN), BPNN optimized by Genetic Algorithm (GA-BP), and BPNN optimized by Artificial Bee Colony Algorithm (ABC-BP). These models were used to forecast six selected time series points. The results indicate that the BPNN model had Mean Absolute Errors (MAE) and Root Mean Squared Errors (RMSE) within 7.6 mm, while the GA-BP model errors were within 3.5 mm, and the ABC-BP model errors were within 3.7 mm. Both optimized models demonstrated significantly improved accuracy and good predictive capabilities.

Monitoring and Prediction of Ground Surface Settlement in Kunming Urban Area by Building GWO-LSTM Model Based on TS-InSAR

Long-term series monitoring of ground surface settlement by remote sensing has become an effective method. Based on TS-InSAR (time series Interferometric Synthetic Aperture Radar) interferometry, this paper proposes a new model based on the Grey Wolf Optimizer (GWO) and Long Short-Term Memory (LSTM) to monitor and predict the ground surface settlement in Kunming City. The results show that the MAPE (mean absolute percentage error), RMSE (root mean square error), and MAE (mean absolute error) of GWO-LSTM are significantly reduced. R2 (goodness of fit) in the six sub-study areas of Kunming City is improved in comparison to the LSTM model. The problem of manual parameter selection in the LSTM model is solved by the GWO algorithm to select parameters automatically. This approach not only significantly reduces the model’s training time but also identifies the most suitable network parameters. This can bring the best performance. Based on TS-InSAR data, the prediction of urban ground surface settlement by the GWO-LSTM model has good accuracy and robustness, which offers a scientific foundation for monitoring and issuing early warnings about urban land disasters.

InSAR-CTPIM-Based 3D Deformation Prediction in Coal Mining Areas of the Baisha Reservoir, China

Time series dynamic prediction of surface deformation in mining areas can provide reference data for coal mine safety and production, which has important impacts. The combination of interferometric synthetic aperture radar (InSAR) technology and the probability integral method (PIM) is commonly used for predicting deformation. However, most surface subsidence prediction in mining areas is based on the static PIM parameters, failing to achieve the three-dimensional (3D) dynamic deformation prediction. This paper proposed a 3D deformation dynamic prediction model (InSAR-3D-CTPIM) between InSAR deformation observations and dynamic coordinate-time PIM (CTPIM) parameters, which can realize the prediction of east–west, north–south, and vertical series deformation caused by mining. The method has been validated by simulation experiments and real experiments in the mining area of Jiansheng Coal Mine in Baisha Reservoir, Henan Province, China. The results showed that the modeling accuracy was improved by 34.3% compared to the traditional multi-rate model, and the accuracy was improved by 28.5% compared to the vertical deformation obtained by the traditional static PIM method. The InSAR-3D-CTPIM model can be used to predict the evolutionary history of basin-wide surface deformation dynamics in coal mining areas, and provide a reference for the early warning and prediction of geological hazards in coal mining areas.

A parameter inversion method for the probability integral method based on robust ridge estimation

The probability integral method is one of the most widely used methods for predicting surface subsidence induced by underground mining in China. In its parameter calculation, the least square algorithm is commonly employed for fitting parameters. However, in the process of fitting parameters, the results are easily affected by ill-conditioned normal matrices and the interference of outliers, resulting in divergent problems. To solve these problems, the principle of robust ridge estimation was introduced in this paper, and a parameter calculation model for the probability integral method based on this principle (hereafter referred to as the established model) was established. Besides, a parameter calculation experiment with manual intervention was conducted in combination with engineering examples. The results demonstrate that the parameter calculation method based on robust ridge estimation can suppress the interference of outliers, overcome the problem of ill-conditioned matrix, and ensure the effectiveness and reliability of parameter estimation results. Compared with the conventional least squares method, the robust ridge estimation method demonstrates greater accuracy in predicting surface subsidence parameters, which validates its rationality and accuracy in underground mining engineering. The research findings provide technical support for obtaining similar parameters for surface subsidence in mining areas and hold significant engineering application value.

Prediction Method for Dynamic Subsidence Basin in Mining Area Based on SBAS-InSAR and Time Function

Dynamic predictions of surface subsidence are crucial for assessing ground damage and protecting surface buildings. Based on Small Baseline Subset Interferometric Synthetic Aperture Radar (SBAS-InSAR) technology, a method for making dynamic predictions of large-scale surface subsidence in mining areas can be established; however, the problem of phase coherence loss in InSAR data makes it impossible to predict the complete dynamic subsidence basin. In this study, a method combining the WeiBull time function and the improved probabilistic integral method (IPIM) model was established based on the PIM model, and a method for predicting the dynamic subsidence basin in the mining area was proposed by integrating the IPIM and the combined WeiBull time function. Time-series subsidence data, obtained using SBAS-InSAR, were used as fitting data, and the parameters of the combined WeiBull function were inverted, pixel by pixel, to predict the dynamic subsidence of the working face in the study area. Based on the predicted surface subsidence results of a certain moment in the working face, the parameters of the IPIM model were inverted to predict the subsidence value in the incoherent region. The subsidence predictions of the combined WeiBull time function and the IPIM model were fused using inverse distance weighting (IDW) interpolation to restore the complete subsidence basin in the mining area. This method was tested at the Wannian Mine in Hebei, and the obtained complete subsidence basin was compared with the measured data, with an absolute error range of 0 to 10 mm. The results show that the dynamic subsidence basin prediction method for the SBAS-InSAR mining area, involving the combination of the IPIM model and the combined WeiBull model, can not only accurately fit the time series of surface observation points affected by mining but also accurately restore the subsidence data in the incoherent region to obtain complete subsidence basin information in the mining area.

Forecasting and early warning of shield tunnelling-induced ground collapse in rock-soil interface mixed ground using multivariate data fusion and Catastrophe Theory

The high spatial variability of the rock-soil interface (RSI) in complex geological conditions introduces strong uncertainties in both subsurface stratigraphy and geotechnical properties. Inaccurate interpretation of such uncertainties during engineering geology investigations increases the geohazard risk of excessive surface settlement or even severe catastrophic ground collapse when shield machines are excavating in RSI mixed ground. Prediction and early warning of excessive surface settlement are necessary measures to address such a risk; however, unavoidable drawbacks such as overfitting, insufficient accuracy, and ineffectiveness remain in existing prediction models and early warning algorithms and have posed significant challenges. In this study, a novel framework using both a multivariate data fusion prediction model and a dynamic early warning algorithm was developed for forecasting and early warning of ground collapse during shield tunnelling in RSI mixed ground. The prediction model is the Differential Evolutionary Optimized Quadratic Taylor Series Extended Kalman Filter (DEQT-EKF); the early warning algorithm is based on Catastrophe Theory and uses the Gradient Ratio (GR) criterion to identify catastrophic singularities. The practicality and accuracy of the framework are well verified by a subway shield tunnelling-induced ground collapse incident in East China with complex RSI mixed ground conditions. The prediction results are compared with the surface settlement measurements and good agreement is obtained, indicating that the DEQT-EKF model can achieve satisfactory accuracy in predicting excessive settlement. The use of the GR criterion can trigger the early warning one time step before the ground collapse event, indicating that it is a competent and practical early warning strategy for shield tunnelling-induced ground collapse. The framework has the potential to significantly reduce the risk of ground collapse caused by geological uncertainties when constructing shield tunnels through complex ground conditions.

Prediction of ground surface settlements by analytical and finite element methods

The soil surrounding the excavation of shallow tunnels may experience destructive deformation near the tunnels. Particularly in densely populated areas where land resources are scarce, ground subsidence caused by tunnel excavation, which in turn affects surrounding structures, deserves more attention. Over the past few decades, various research methods for predicting the maximum vertical ground surface settlement have been proposed. In this study, an empirical estimation method, an analytical technique for evaluating ground movements, and a numerical simulation method using Plaxis 2D, a finite element model software, were selected as prediction approaches for ground surface settlements. Subsidence measurements were obtained from a metro tunnel excavated using an earth pressure balance shield tunnel boring machine and constructed in a dense sandy gravel layer. Comparison of the monitored deformations with the estimation results calculated by different prediction approaches showed that the maximum vertical settlement of the ground surface evaluated by the analytical technique and finite element model agreed well with the actual measured values. This suggests that these two prediction methods have the potential for conserving nearby structures during the construction of underground infrastructures.

Prediction of airport runway settlement using an integrated SBAS-InSAR and BP-EnKF approach

The prediction of surface settlement occupies a crucial role in achieving effective catastrophe prevention and mitigation, as well as facilitating the maintenance of airport runways. Given the challenges associated with the manual collection of settlement data and the suboptimal timeliness of the Back Propagation (BP) technique in neural networks, this study proposes an integrated prediction method that combines the Ensemble Kalman Filter (EnKF) with BP. This study focuses on the processing of Synthetic Aperture Radar (SAR) pictures obtained from the ascending orbit of the Sentinel-1A satellite. The Small Baseline Subset Interferometric synthetic aperture radar (SBAS-InSAR) technology is employed to derive the settlement time series on the runway of Kangding Airport. Moreover, three sites with high coherence within the primary settlement region are employed to assess the dependability of the model after the expansion of the data by cubic spline interpolation. The findings of the study indicate that both the BP-EnKF and BP models exhibit favorable outcomes in predicting airport runway settlement. However, following the alteration of data caused by external environmental influences, the BP-EnKF model has superior adaptability to variations in data. It has been shown that the BP-EnKF model exhibits a prediction accuracy that surpasses the BP model by 9.25%.

Application of a Variable Weight Time Function Combined Model in Surface Subsidence Prediction in Goaf Area: A Case Study in China

To attain precise forecasts of surface displacements and deformations in goaf areas (a void or cavity that remains underground after the extraction of mineral resources) following coal extraction, this study based on the limitations of individual time function models, conducted a thorough analysis of how the parameters of the model impact subsidence curves. Parameter estimation was conducted using the trust-region reflective algorithm (TRF), and the time function models were identified. Then we utilized a combined model approach and introduced the sliding window mechanism to assign variable weights to the model. Based on this, the combined model was used for prediction, followed by the application of this composite prediction to engineering scenarios for the dynamic forecasting of surface movements and deformations. The results indicated that, in comparison with DE, GA, PSO algorithms, the TRF exhibited superior stability and convergence. The parameter models obtained using this method demonstrated a higher level of predictive accuracy. Moreover, the predictive precision of the variable-weight time function combined model surpassed that of corresponding individual time function models. When employing six different variable-weight combination prediction models for point C22, the Weibull-MMF model demonstrated the most favorable fitting performance, featuring a root mean square error (RMSE) of 32.98 mm, a mean absolute error (MAE) of 25.66 mm, a mean absolute percentage error (MAPE) of 7.67%; the correlation coefficient R2 reached 0.99937. These metrics consistently outperformed their respective individual time function models. Additionally, in the validation process of the combined model at point C16, the residuals were notably smaller than those of individual models. This reaffirmed the accuracy and reliability of the proposed variable-weight combined model. Given that the variable-weight combination model was an evolution from individual time function models, its applicability extends to a broader range, offering valuable guidance for the dynamic prediction of surface movement and deformation in mining areas.