[Objective] The steel volute of a pumped storage power station is the part of the flow passage structure subjected to the highest internal pressure, bearing cyclic water pressure during operation and facing potential risk of low-cycle fatigue failure. At present, a fundamental issue in predicting low-cycle fatigue life of steel volutes in pumped storage power stations lies in the scientific determination and input of the fatigue load spectrum. [Methods] Static analysis of the composite structure was performed on the Abaqus finite element platform. Based on water level monitoring data, static monitoring results, and unit operating modes, the prototype load spectrum, rainflow-counting load spectrum, and constant-amplitude load spectrum were respectively compiled. [Results] By comparing with the fatigue life prediction results based on the prototype load spectrum, the reliability of the load spectrum compiled by the rainflow-counting method was verified. The sequence of cycle amplitudes and extremely small amplitude loads in rainflow counting had minimal impact on fatigue life prediction results, indicating that the rainflow-counting method could serve as a simplified input approach for low-cycle fatigue loading of steel volutes. The fatigue life prediction results based on the constant-amplitude load spectrum were close to those based on the prototype spectrum. For this pumped storage power station, constant-amplitude loading could be used as a simplified input for fatigue loading. However, it should be noted that the drawdown depth of the water level at this power station was relatively small. In such cases, whether the prototype load spectrum, rainflow-counting load spectrum, or constant-amplitude load spectrum was used, the range of cyclic amplitude variation was limited. For power stations with small water level drawdown depths, the fatigue load spectrum of the steel volute could be simplified to a constant-amplitude form. However, for power stations with relatively large water level drawdown depths, whether the above conclusions were applicable required further investigation. [Conclusion] The findings of this study can provide a reference for compiling low-cycle fatigue load spectra for steel volutes. According to the prediction results, there is no risk of low-cycle fatigue failure during the operation period of the power station. However, certain limitations in the calculations of this study may lead to an overestimation of prediction results for the following three reasons. (1) Only the impact of hydrostatic pressure on the steel volute is considered, while variations in water hammer pressure during transitions of unit operational states are not taken into account. (2) Seasonal variations in water temperature inside the steel volute significantly affect the timing and spatial distribution of contact closure between the steel volute and concrete, and neglecting temperature effects may underestimate the stress level in the steel volute. (3) During operation, the steel volute and concrete jointly bear the internal water pressure. Cracking in the concrete weakens its restraining effect on the steel volute, leading to an underestimation of the stress level in the steel volute in calculations.

[Objective] Dam deformation is comprehensively influenced by multiple components such as water level, temperature, and time-dependent effects, exhibiting characteristics of nonlinear time series. Currently, traditional and single models struggle to fully capture the complexity and diversity of dam deformation data, resulting in limited predictive performance and interpretation ability. To solve the above problems, this study aims to propose an efficient and interpretable dam deformation prediction method through the combination and optimization of multiple prediction models. [Methods] First, the least absolute shrinkage and selection operator (LASSO) was used to efficiently screen numerous environmental variables, both simplifying model input and explaining the reliability of factor selection.Then,the long short-term memory (LSTM) network was employed to predict dam deformation, and the attention mechanism was introduced to enhance the extraction of important information.Finally,the bagging algorithm was used to integrate the prediction results of multiple models, further improving the accuracy, stability, and generalization ability of the overall prediction. By combining the advantages of LASSO regression feature selection, LSTM model with attention mechanism, and bagging ensemble algorithm, a multi-model coupled method was proposed. [Results] To validate the effectiveness and applicability of the coupled model, this study took the deformation monitoring data of a roller-compacted concrete gravity dam as the research object for prediction analysis. When the number of features was relatively large, the LASSO variable selection method reduced model complexity by adding L1 regularization term and selected features with important influence on dam displacement, enhancing the interpretability of factor selection. Combined with this method, multiple LSTM models with attention mechanism were integrated for parallel training and prediction, reducing potential overfitting problems in single models and improving generalization ability and prediction efficiency of the coupled model. The trained model was used to predict and validate the test set data. The residual values of the coupled model were small, and residual distribution had strong randomness, indicating high prediction accuracy. The fitting results of each measurement point were smooth and agreed well with the measured data, and the prediction results were stable without showing any “distortion” phenomenon. Using the same dataset and identical proportion division, LSTM multi-factor model, stepwise regression prediction model, LASSO regression model, LASSO-LSTM model, and the coupled model were compared and analyzed. The results showed that the coupled model proposed in this study significantly outperformed other models in overall prediction trend and the prediction accuracy of partial fluctuations. The average MAE, MSE, and RMSE at each measurement point were 0.052, 0.005,0.067 mm, respectively. The coupled model could more accurately capture the dynamic changes of dam deformation, providing a simple and efficient method for prediction model research. [Conclusion] This study constructs a coupled prediction model with high accuracy, stability, and interpretability. The main innovation lies in the effective selection of key environmental variables through LASSO, simplifying model input and improving its interpretability; the use of LSTM to capture the time-series features of dam displacement data, while the incorporated attention mechanism helps the model focus on important features in time series; and the bagging algorithm that significantly improves the generalization ability of model by training multiple sub-models in parallel. Based on actual case analysis, the coupled model not only demonstrates higher accuracy in dam deformation prediction,but also outperforms commonly used models in interpretability and stability.The coupled model based on interpretable variable selection provides a reference for the optimization of subsequent combined models.Future research directions can shift from single measurement points in different dam sections to multiple measurement points in the same dam section.This will involve analyzing the location of measurement points and the relationships between different points,thereby enabling the construction of a comprehensive multi-point coupled model.

[Objective] This study aims to systematically investigate the control measures for geyser intensity in baffle-drop shafts during the release of high-pressure trapped air pockets. By analyzing the effects of key parameters—including the connection mode of the communication pipe, the area of the connecting region between dry and wet zones, the position and open area of the throttling orifice plate, and the installation distance of the vent pipe—on the geyser height and the impact load on baffles, a set of comprehensive optimization measures balancing geyser control effectiveness and structural safety is proposed. [Methods] FLUENT software was used to establish a three-dimensional numerical model of geyser in a baffle-drop shaft based on the Realizable k-ε turbulence model and the VOF two-phase flow model. The Dongfeng Road baffle-drop shaft of the Donghao Chong deep tunnel project in Guangzhou was selected as the research object. The effects of different communication pipe connection modes (dry zone/wet zone), areas of the connecting region between dry and wet zones, throttling orifice plate parameters (height and open area), and vent pipe installation distance on the jet height of geysers and the impact load on baffles were systematically simulated. A total of 88 working conditions were simulated, and model reliability and computational accuracy were ensured through grid independence verification and comparison with experimental data. The response patterns of geyser intensity and baffle impact load to each parameter were analyzed in detail. [Results] Although connecting the communication pipe to the wet zone had a limited effect on the geyser height, it significantly reduced the impact load on the baffles—particularly on the bottom baffle, where the peak load was reduced by up to 66%. The area of the connecting region between the dry and wet zones showed a nonlinear relationship with the baffle load; as the area decreased, the impact load on the bottom baffle increased markedly. The optimal control effect was achieved when the dimensionless area S^*=0.318. When the throttling orifice plate was positioned at the mid-height of the dry zone (1/2H) with an open area of ϕ^*=0.058, the maximum geyser height decreased by approximately 70%, while the impact load on the baffles dropped by more than 30%. The best control effect was achieved when the vent pipe was positioned at the end of the communication pipe farthest from the shaft (δ^*=4D), and no water-air mixture overflow occurred. [Conclusion] Considering the combined influence of these factors on the geyser intensity in the shaft, a joint control measure—“wet-zone connection + mid-position throttling orifice plate + remote vent pipe + optimized connecting area layout”—was proposed. Under typical working conditions, this combined approach reduced the geyser height by up to 80% and the average impact load on the baffles by more than 50%, effectively controlling the geyser intensity while ensuring the structural safety of the shaft. It provides reliable theoretical support and practical guidance for the safe design and operation of baffle-drop shafts in deep tunnel drainage systems and offers a replicable and scalable technical approach for future geyser risk prevention and control in deep tunnel systems.

[Objective] The 3D laser scanning technology is characterized by fast scanning speed, high scanning accuracy, non-contact operation, and minimal influence from the scanning environment, which makes it widely applicable in deep engineering fields. However, high in-situ stress and complex geological structures result in complex tunnel surface morphology and a non-linear actual axis, making the filtering and classification of point clouds for deeply buried tunnels more difficult than those for shallow-buried projects. This study aims to address the recognition and classification of point cloud profile for deeply buried irregular tunnels. [Methods] Based on the spatial morphology of the contour of deeply buried irregular tunnel, we established a two-level filtering method for the point cloud of deeply buried irregular tunnels, and developed a tunnel point cloud classification method based on density-based clustering and spatial position classification. [Results] To verify the effectiveness of the proposed methods, the point cloud data of a 30 m-long deep tunnel excavated by the drilling and blasting method were used as the research object. First, two 1 m-long tunnel segment point clouds were selected for analysis. The filtering effects of the tunnel segment point clouds were analyzed under different segment thicknesses L=0.2, 0.4, 0.6 m and distance thresholds d_critical=0.02, 0.04, 0.06, 0.08,0.1 m. Through comprehensive comparison, the optimal parameters were determined as L=0.2 m and d_critical=0.04 m, which were successfully applied in the filtering of the tunnel point cloud. On this basis, according to the spatial distribution characteristics of the tunnel segment point clouds, the DBSCAN algorithm parameters were set to ε=0.1 m and MinPts=50, which enabled the classification of non-profile point clouds of tunnel segments. [Conclusion] This study focuses on the filtering problem of point clouds in deeply buried tunnels. Based on the spatial geometric features of tunnel point clouds, a two-level filtering and classification method for point clouds of deeply buried tunnels with complex morphology is proposed. Case analysis shows that the proposed method realizes effective filtering and classification of tunnel point clouds with complex morphology, solves the recognition problem of contour point clouds in deeply buried tunnels, and provides reliable technical support for applications of 3D laser scanning point clouds in potential risk area identification, lining thickness detection, spatiotemporal deformation monitoring, and health condition assessment of deeply buried tunnels.

Hydraulic concrete structures are prone to various types of defects during construction and service. These defects are primarily categorized into three major types: apparent defects, internal defects, and defects at structure-foundation connections. Apparent defects, such as surface cracks, spalling, and cavitation, are primarily induced by thermal stress, shrinkage deformation, and environmental erosion. Specifically, the scouring and abrasion from sediment-laden flow can lead to surface spalling, while high-velocity water flow tends to cause cavitation. Internal defects mainly manifest as honeycombs and voids, which are primarily caused by issues like entrapped air bubbles and grout leakage from formwork due to the difficulties in underwater vibration. These issues are particularly prone to occur in areas with dense reinforcement or in mass concrete. Defects at structure-foundation connections primarily include misalignment and differential settlement, mainly resulting from the combined effects of multiple complex factors, such as repeated hydraulic pressure, uneven foundation settlement, and temperature variations. For concealed and hard-to-access underwater structural defects, non-destructive testing (NDT) technologies demonstrate distinct advantages. This study systematically reviews the research progress on four major NDT methods: optical imaging, sonar scanning, sub-bottom profiling, and impact-echo method. Optical imaging can effectively identify apparent defects through image analysis. However, affected by the optical properties of water, it suffers from problems such as poor image quality and limited identification accuracy. Sonar scanning can overcome the limitations of turbid water and achieve large-scale detection. However, its imaging resolution is relatively low, and it lacks a systematic correspondence between defect features and image features. Sub-bottom profiling, based on the strong penetration capability of low-frequency sound waves, shows potential in detecting internal defects of underwater structures and foundation conditions. However, its application research in the hydraulic engineering field remains relatively limited. The impact-echo method enables the detection of internal defects by analyzing the propagation characteristics of stress waves, unaffected by water and steel reinforcement. However, it still faces challenges in signal interpretation and quantitative evaluation. Based on the analysis and discussion of current research on NDT technologies for underwater structural defects, future development of these technologies should focus on the following directions: (1) establishing a deep learning-driven multi-source data fusion framework to enhance the capability of defect feature recognition; (2) developing opto-acoustic collaborative detection technologies that integrate the detailed resolution capability of optical imaging with the environmental adaptability of sonar; (3) developing more advanced stress-wave signal processing algorithms and quantitative evaluation models to improve the detection accuracy of the impact-echo method. Through multi-technology integration and intelligent development, it is expected that more comprehensive and accurate detection and assessment of underwater structural defects can be achieved, thereby providing strong technical support for the safe operation of hydraulic engineering projects.

[Objective] Dynamic groundwater changes represent one of the key controlling factors in the initiation of soil landslides. Investigating their response characteristics and mechanisms under rainfall is crucial for understanding landslide stability and evolutionary processes. [Methods] Taking the typical thick soil landslide—Tanjiawan landslide—in the Three Gorges Reservoir area as the research subject, this study systematically analysed the influence mechanism of groundwater level dynamics under rainfall by relying on multi-year continuous high-precision GNSS surface displacement data, automated groundwater level monitoring data, regional rainfall records, and information obtained from repeated field geological surveys on landslide geological structure, sliding mass structure characteristics, and groundwater recharge and discharge conditions. [Results] Deformation of the Tanjiawan landslide was concentrated in the mid-front and left-side areas and was closely related to rainfall events. Antecedent cumulative rainfall, to a certain extent, determined the slope's deformation response to a subsequent single heavy rainfall event. When cumulative rainfall was sufficient, even moderate single rainfall intensity may induce significant deformation. Groundwater level changes and rainfall infiltration showed distinct spatiotemporal correlation. The rate of water level rise was influenced by rainfall infiltration conditions and jointly controlled by both the antecedent effect and the concurrent effect of rainfall intensity. After rainfall infiltration, dynamic groundwater migration followed two main paths: one along route AB rapidly converged on the frontal area, generating strong hydrodynamic pressure; the other migrated slowly within the sliding mass, producing a cumulative effect on overall water content and pore water pressure. The hydrodynamic pressure generated along route AB, when coupled with local topographic conditions, directly drove deformation of the I-1 sliding mass, triggering local accelerated deformation or even failure. [Conclusions] Slope deformation trends are significantly controlled by dynamic groundwater changes, whereas slope stability is, to a certain extent, constrained by the duration of peak groundwater level. Prolonged high water levels markedly reduce slope stability. Moreover, monitoring data shows a certain lag between groundwater level rise and slope deformation rate, a characteristic that provides important reference value for early identification and early warning of thick soil landslides. This study provides a theoretical basis for research on the deformation mechanism, early identification, and early warning of soil landslides under rainfall conditions.

[Objective] In underwater engineering inspection, the turbid shallow water environment severely hinders the performance of machine vision-based methods for detecting surface defects in underwater structures. To address the challenge of defect detection in turbid water, this study proposes a lightweight three-stage underwater defect detection method that integrates polarization imaging and deep learning techniques. A defect detection model, named PCC-YOLOv7, is developed. [Methods] First, polarization imaging technology was combined with a polarization restoration model to analyze the polarization characteristics of light waves. This approach effectively suppressed scattering interference in turbid water, thereby achieving clear imaging of turbid environments and restoring defect images. Consequently, defect details obscured by scattering particles were reconstructed. Second, the CAA-SRGAN (Coordinate Attention ACON-Super Resolution Generative Adversarial Network) model was introduced. By employing an improved attention mechanism and a generative adversarial network structure, super-resolution processing was performed on the restored images. This yielded high-resolution underwater defect images, providing a high-quality data foundation for subsequent precise detection. Finally, a defect detection model based on CBAM-YOLOv7 was established, where the convolutional block attention module (CBAM) was utilized to enhance the network’s focus on defect features. Leveraging the advanced YOLOv7 object detection framework, common underwater structural defects, including cracks, holes, and spalling can be rapidly and accurately identified. These three sub-models worked collaboratively to form a comprehensive detection system. [Results] For image restoration, the polarization restoration model exhibited superior performance in metrics such as image clarity and color fidelity compared to current restoration methods. The CAA-SRGAN model generated images with notable improvements in detail texture preservation and resolution enhancement. The CBAM-YOLOv7 defect detection model achieved higher accuracy in both defect localization and classification. A comprehensive evaluation of the PCC-YOLOv7 defect detection model revealed an average improvement of 33.5% in mean average precision (mAP_0.5, mAP_0.75, and mAP_0.5-0.95). Compared to existing models, PCC-YOLOv7 significantly enhanced defect detection performance in turbid underwater environments, effectively improving both recognition rate and detection efficiency. [Conclusions] The PCC-YOLOv7 defect detection model innovatively integrates polarization imaging technology with deep learning. Through the collaborative operation of three functionally complementary sub-models, it successfully addresses the challenge of detecting surface defects in underwater structures in turbid water. Compared to existing models, the proposed model demonstrates enhanced adaptability to turbid underwater detection scenarios. It enables stable and efficient detection of surface defects in underwater structures under complex turbid conditions, providing a practical technical solution for the safety assessment and maintenance of underwater structures. Future work may focus on further optimizing the model structure and extending its application to more underwater scenarios.

[Objective] The material composition of high core rockfill dams is complex. Under the influence of hydraulic loading, temperature, and other complex environmental factors during long-term service, the permeability coefficients of these materials inherently exhibit time-varying characteristics, significantly influencing seepage stability and overall dam safety. This study aims to address the limitations of existing research, which predominantly focuses on static parameter inversion or non-time-varying risk assessment, and lacks systematic consideration of the time-varying patterns of material parameters and the influence of long-term operation. [Methods] A time-varying risk analysis method for seepage failure in high core rockfill dams was proposed, integrating data decomposition, finite element simulation, time-varying parameter inversion, and failure risk analysis. First, based on multi-year measured seepage pressure data of the dam, empirical mode decomposition was used to extract the periodic and trend components. Combined with orthogonal experiments and response surface methodology, an inverse surrogate model for the permeability coefficients of the high core rockfill dam and its foundation was constructed. Subsequently, through optimization using a genetic algorithm, this study investigated and revealed the time-varying patterns and characterization functions of permeability coefficients. Finally, based on the time-varying patterns of permeability coefficients and the Monte Carlo method, a time-varying risk analysis model for seepage failure in high core rockfill dams was established to achieve the dynamic risk assessment of seepage failure in the dam structure. [Results] This method was applied to the Pubugou Dam project. The results showed that the relative error of permeability coefficient inversion for both the dam and foundation was less than 2%, with an average relative error of 0.4%. Seepage field simulations based on the inverted parameters showed that the distribution of seepage pressure inside the dam followed the rising and falling trend of the reservoir water level and was consistent with the patterns observed in the measured seepage pressure data. This conformed to the typical seepage field distribution patterns of high core rockfill dams, indicating a high level of inversion accuracy. Furthermore, the permeability coefficients of both the core wall and the overburden layer showed time-varying patterns of gradual increase and stabilization. Reliability analysis of seepage failure in the dam and its foundation indicated that the reliability indicator (β) of the dam consistently exceeded the design target value during the operational period, suggesting that the overall risk of seepage failure was low. Additionally, the reliability indicator for seepage failure in the dam and its foundation exhibited periodic fluctuations with changes in the reservoir water level, showing a generally negative correlation between the reliability indicator and reservoir water levels and a positive correlation between failure probabilities and water levels. This was generally consistent with the seepage characteristics and patterns of dam structures under different water level conditions, validating the applicability of the proposed time-varying risk analysis model. The results confirmed that the reliability indicator for seepage failure of the Pubugou Dam complied with regulatory requirements. [Conclusions] The method developed in this study integrates time-varying parameter inversion, modeling of time-varying patterns, failure path search, surrogate model construction, and time-varying risk analysis. By dynamically identifying and updating time-varying parameters in real time, it enables accurate simulation of seepage failure processes and full lifecycle monitoring of risk evolution, thereby enhancing the timeliness and accuracy of safety risk assessments for high dam structures.

[Objective] The attitude of a shield machine is a critical parameter that significantly affects tunnel construction, directly determining construction safety and project quality. To ensure that shield tunneling closely aligns with the designed alignment and to improve engineering construction quality, this study proposes a novel shield attitude prediction model, called WM-CTA, based on deep learning technology. [Methods] The WM-CTA model primarily consists of two frameworks: a data preprocessing module (Wavelet Transform and Maximum Information Coefficient) and a prediction module (Convolutional Neural Network and Attention Mechanism). The preprocessing module, composed of Wavelet Transform (WT) and the Maximum Information Coefficient (MIC) algorithms, was used to perform noise reduction and parameter correlation analysis on the raw data, thereby generating enhanced inputs. The Convolutional Neural Network (CNN) integrated with a channel-wise attention mechanism explored parameter weight differences and extracted local data features. Subsequently, the Temporal Convolutional Network (TCN) was employed to capture temporal dependencies and dynamic variations in the data. Finally, the Attention Mechanism (AM) was applied to extract key temporal node information. The model’s prediction performance was validated using monitoring data from a section of a shield tunnel under construction in Shenyang. Experiments were conducted on data for noise reduction and correlation analysis, followed by analysis of the model’s prediction performance and generalization ability. [Results] Experimental results showed that the monitoring curves processed with wavelet transform had improved smoothness with reduced frequency of abrupt changes between data points. Correlation analysis indicated that shield construction parameters exerted greater influence on shield attitude than soil parameters, enabling dimensionality reduction of input parameters. Compared with four baseline models, the proposed WM-CTA model achieved minimum MAE and RMSE and maximum R² value. [Conclusion] The experiments verify that the WM-CTA model delivers optimal prediction performance with high computational efficiency. Furthermore, the model exhibits strong generalization ability, providing valuable references for similar future engineering projects.

[Objective] Dam deformation results from the nonlinear effects of multiple complex environmental factors. Traditional mathematical models for dam deformation monitoring have difficulty reflecting the complex nonlinear relationships between effect variables and environmental variables, often leading to unsatisfactory prediction results. By leveraging the long-short-term memory (LSTM) model and particle swarm optimization (PSO) algorithm from artificial intelligence technology, a combined PSO-LSTM dam deformation prediction model is established, offering a novel approach for enhancing the accuracy of dam deformation prediction. [Methods] By applying PSO for global optimization of LSTM hyperparameters, a combined PSO-LSTM dam deformation prediction model was established. This method both addressed the deficiencies of traditional prediction models in describing nonlinearity between variables and enhanced the appropriateness of LSTM hyperparameter values. The specific methods included: constructing environmental variable factors based on the interaction mechanism between dam deformation and environmental variables; inputting deformation training sets to determine the range of hyperparameters to be optimized and training the network hyperparameters using the LSTM model; setting the particle position information as the hyperparameters to be optimized and using the PSO algorithm to optimize the LSTM hyperparameters; and outputting dam deformation predicted values at different prediction time points using the parameters obtained from training. [Results] Utilizing deformation monitoring data from concrete gravity dams and concrete arch dams, this study established a traditional monitoring statistical model, a standalone LSTM prediction model, and a combined PSO-LSTM model. The results showed that: (1) the combined PSO-LSTM model achieved the smallest RMSE and MAE values and the largest R² value, indicating excellent prediction accuracy. Compared to statistical models for monitoring and standalone LSTM models, it demonstrated significantly improved prediction performance. (2) Due to its strong nonlinear learning capabilities, the combined PSO-LSTM model could effectively extract nonlinear characteristics from complex datasets, thereby achieving good prediction performance even with poor-quality deformation monitoring data. [Conclusion] (1) The combined prediction model established based on LSTM and PSO algorithms effectively extracts nonlinear characteristics between environmental variables and effect variables, leading to improved prediction performance. (2) The PSO-LSTM prediction model demonstrates good versatility. Its fundamental principles apply not only to concrete dams but also to earth-rock dams and other hydraulic engineering projects. However, when applying the model, the configuration of neurons in the LSTM model’s input layer must be tailored to the structural characteristics, operational conditions, and influencing factors of different dam types.

[Objective] Since the impoundment of the Three Gorges Dam in 2003, many cataclastic bedding landslides in the reservoir area have been reactivated due to the influence of external factors such as reservoir water level fluctuations, variations in groundwater levels, and rainfall. These landslides are typically large in scale, exhibit complex deformation mechanisms, and pose significant challenges for early warning and disaster prevention. This paper attempts to establish an interaction model linking rainfall, reservoir water, and groundwater with groundwater as the main triggering factor of landslide deformation, and to further reveal the dynamic migration patterns of groundwater under the coupled effects of reservoir water level fluctuations and rainfall. [Methods] The study took the Muyubao Landslide, a cataclastic bedding landslide in the Three Gorges Reservoir area, as a case study. By integrating data statistics with field investigations, a statistical analysis was conducted on nearly seven years of manual and automated monitoring data, field investigation materials, and hydrometeorological information to investigate the quantitative relationship among reservoir water level, rainfall, and groundwater level, as well as the interaction between groundwater level and slope displacement and deformation. [Results] The research results indicated that: (1) A groundwater level of 175 meters at the front platform of the slope served as the critical threshold for the initiation of landslide deformation. When the groundwater level at the front platform approached 175 m, deformation began under the influence of buoyancy-induced weight reduction. When the groundwater level at the front platform exceeded 175 m, both buoyancy-induced reduction and dynamic water pressure acted on the slope, with the effect of dynamic water pressure intensifying as the groundwater level rose.(2) Statistical analysis of monitoring data revealed thresholds for rainfall-induced groundwater level rise. Ten consecutive days of rainfall totaling 150 mm increased the groundwater level at the front of the slope by 3.22 m to 6.88 m. In the case of 300 mm of cumulative rainfall within 30 days, the groundwater level at the front of the slope increased by approximately 10 m. A “lag effect” was observed in groundwater response to rainfall, typically lasting 3 to 13 days.(3) From October to December 2017, rainfall occurred on 27 out of first 32 days, totaling 310.6 mm. As a result, the groundwater level in borehole QSK1 at the front platform of the slope rose to 184.2 m, nearly 10 m above the highest reservoir water level (175 m). Over the 72-day period, the slope displacement totaled 88.5 mm. [Conclusion] (1) Groundwater level fluctuations significantly precede slope deformation. Given known reservoir water levels, it is possible to forecast groundwater level based on reservoir water level and rainfall data, and further predict slope deformation based on the groundwater level. This approach provides a strong basis for landslide early warning and prediction.(2) The effective contribution of rainfall to groundwater recharge varies with different types of rainfall. Compared to intense rainstorms, prolonged and continuous rainfall causes a more significant rise in groundwater levels, especially around periods of high reservoir water levels, posing greater risks to slope stability. Therefore, in landslide disaster prevention, the role of rainfall and groundwater should be carefully considered. It is crucial to optimize the layout of monitoring points, enhance real-time automated groundwater monitoring capacity and service quality, and better understand the impact of groundwater dynamics on slope deformation.

An intelligent prediction model for shield tunneling parameters accounting for geological conditions is presented by employing a geological data processing technique which integrates in-situ stress extraction and tunneling section stratum information coding. The method encompasses data acquisition, preprocessing and decomposition, and model construction, training and testing, as well as result evaluation and analysis. The model is applied to predict the shield parameters for the large-diameter slurry shield tunnel project of the Luyuan North Street section on the Beijing-Harbin Expressway within the Beijing East Sixth Ring Road reconstruction project. Findings reveal that accounting for geological conditions enhances the prediction accuracy of shield thrust and cutter-head torque by 38.53% and 44.86%, respectively. This improvement secures the construction safety of subsequent shield tunneling operations. The research outcomes can serve as a reference for future similar projects.

Continuous slope monitoring in reservoirs and dams using ground-based synthetic aperture radar interferometry (GB-InSAR) is vulnerable to atmospheric environmental fluctuations. These fluctuations can cause inaccuracies in deformation results derived from interferogram sequences. Moreover, processing large volumes of continuous GB-SAR images is time-consuming, which negatively affects the overall efficiency of GB-InSAR and the feasibility of quasi-real-time deformation analysis applications. To tackle these problems, this paper presents a uniform grid sampling method and interferometric stacking technique based on the phase gradient building on the conventional polynomial atmospheric correction method. A polynomial atmospheric correction method based on downsampled high-quality pixels (HQPs) is then constructed. This method is applied to monitor the deformation of the high slope on the right bank during the construction of the Huangdeng Hydropower Station. Experimental results show that the root mean square error (RMSE) of the binary polynomial model averages 0.039 5 rad, significantly outperforming that of the unitary model and other conventional correction methods. The average RMSE of the proposed method is 0.024 0 rad, comparable to the accuracy before downsampling. However, the overall solution time reduces notably from 2.32 h to 0.80 h. This indicates that the proposed method can significantly improve the efficiency of continuous image atmospheric correction while maintaining modeling accuracy, offering effective technical support for slope safety monitoring.

The service time and time-varying characteristics of safety monitoring instruments in hydropower projects are critical factors affecting the design, implementation, and operational management of engineering safety monitoring. However, there is limited research on the time-varying behavior of these instruments both in China and abroad. In this study we collected actual large-sample data on the service performance of safety monitoring instruments from ten hydropower projects in China. Important time nodes for the ten projects were summarized, and the total failure rates of monitoring instruments in each project were calculated, revealing the time-varying patterns of failure rates. Results indicate that: 1)During the construction phase, the failure rate increment model of monitoring instruments in seven projects follows a normal distribution, with the failure rate increment in six projects initially increasing and then decreasing. The inflection point occurs at 51%-84% of the construction period. The failure rate of monitoring instruments in three projects consistently increased throughout the construction phase. 2)During the operational phase, the failure rates of monitoring instruments exhibited a steady upward trend, conforming to a linear model. The time-varying models proposed in this paper effectively predict the changing failure rates of safety monitoring instruments in hydropower projects.

To investigate the deformation characteristics of retaining piles in the footwall influenced by normal faults, a case study of a foundation pit project in Shenzhen City was conducted using a comprehensive approach that included numerical simulations and field measurements. The study examined how different fault slip amounts, dip angles, and positions affect the deformation of retaining piles in the footwall’s influence zone. Sensitivity analysis and orthogonal experiments were carried out to assess the impact of these fault parameters. Results revealed that deformation of the retaining piles decreased under the normal fault, with the center of gravity shifting downward. The upper sections of the piles experienced more significant deformation compared to the lower sections. Deformation was inversely proportional to both the fault slip amount and dip angle, and directly proportional to the distance from the fault to the foundation pit. Specifically, the maximum deformation rate, r(ΔZ_max/Δ), decreased exponentially with increasing fault slip amount and dip angle, but increased logarithmically with increasing distance from the fault. Sensitivity analysis showed that dip angle had the most significant impact on the maximum deformation of the retaining piles, followed by slip amount, with the fault position having the least influence. By fitting data from 64 orthogonal experiments, a strong linear relationship was established between the maximum deformation U_hm and the index η(\frac{\theta \mathrm{\pi }T}{180°S}).Consequently,a predictive model for the maximum deformation of retaining piles in the footwall’s influence zone was developed, along with a corresponding predictive equation for this project. These findings offer valuable insights for deformation control in foundation pit projects located in normal fault areas with similar geological conditions.

Traditional single-model prediction methods suffer from issues like low accuracy, susceptibility to noise, and limited generalization capability. To address these challenges, we propose a novel approach for predicting concrete dam deformation by integrating the Beta Prior Principal Component Analysis (BP-PCA) and the Water Cycle Algorithm (WCA). Initially, the BP-PCA model decomposes deformation data into multiple scales, effectively reducing noise. This decomposition transforms the intricate nonlinear and non-stationary stochastic process into a set of principal components with simplified structures. Simultaneously, it enhances noise robustness by suppressing noise during the decomposition process. Subsequently, we employ the Water Cycle Algorithm optimized Support Vector Machine (WCA-SVM) to construct prediction models for each principal component. Finally, we integrate the prediction outcomes from multiple principal components to derive the final prediction result. The relative prediction error is minimized to 1.07%, with a root mean square error of 0.065. Compared to the three methods included in the comparative analysis, our approach yields over 62% improvement in prediction performance, demonstrating superior noise robustness and generalization capability.

The step-like evolution of landslide represents the fluctuating behavior of landslides influenced by external factors during the isokinetic deformation phase. Step-like landslide is characterized by extended deformation cycles, intricate mechanisms, and challenges in disaster early-warning. By analyzing the deformation-time curves of step-like landslide, we introduced the concept of “one rainfall process” and defined multiple rainfall intervals in the monitoring sequence. Subsequently, we categorized the warning process into two modes: the previous rainfall-plus the current rainfall pattern, and the current rainfall pattern. With cause-time-space as significant indices for landslide warning, we established a holistic landslide early-warning criterion model, and designed a dynamic early-warning system for Landslide No. 1 at Machi Village as a case study. By correlating geological conditions and monitoring data with a profound analysis of deformation evolution patterns and warning thresholds, we observed that: 1) The landslide deformation mode is predominantly traction-related, demonstrating a typical rainfall-triggered step-like behavior. 2) The effective early rainfall duration is 10 days. The rainfall thresholds are 24 mm and 32 mm respectively under the previous plus current rainfall mode, and 37 mm under the current rainfall mode. 3) With the threshold values for rainfall and displacement rate as the Grade III yellow early-warning central boundary, we established a comprehensive dynamic grading early-warning system that transitions from traditional threshold warning to process warning. This shift enhances the precision and efficiency of landslide prediction and management.

The aim of this research is to explore the impact of the 6.8 magnitude earthquake in Luding County,Ganzi Prefecture of Sichuan Province on Dagangshan High Arch Dam on September 5,2022. According to the monitoring data of 21 earthquake sensors installed across the dam body and its foundation,we analyzed the time and frequency domains of the seismic recordings from the Luding earthquake to unravel the dynamic response patterns of Dagangshan High Arch Dam during the earthquake. Findings reveal substantial displacements and accelerations along the dam crest and the abutments on both sides. Notably,the highest recorded acceleration peak,reaching 576.6 cm/s²,was found at the No.6 dam section atop the crest. The seismic impact on Dagangshan High Arch Dam manifests prominently within the frequency spectrum of 0.5 to 8 Hz. Moreover,the dam’s towering stature accentuates ground motion response,particularly amplifying peak accelerations along the riverbanks. Post-earthquake assessment indicates the dam’s operational stability with minimal impact on its overall integrity. However,vigilance is warranted for the dam abutments and adjacent slopes,necessitating meticulous observation and maintenance measures.

In addressing the substantial data volume within landslide monitoring databases and the lengthy processing times due to multiple database scans required for association rule analysis, we introduce the Eclat association rule algorithm into landslide monitoring data mining. This approach involves analyzing the deformation of the Bazimen landslide using the K-means clustering method and the Eclat algorithm. Through comprehensive investigation, we identify six factors from rainfall monitoring values and reservoir water level monitoring values for data mining and analysis. By uncovering the correlations of three rainfall factors and three reservoir water level factors with the displacement of multiple measurement points in the Bazimen landslide, we extract eight association rules with a high confidence level from all excavated correlation rules derived from the spatiotemporal monitoring big data of the Bazimen landslide. This analysis reveals effective information of rainfall and water level influencing landslide movement. The findings indicate the potential widespread applicability of this data mining method due to its high accuracy in monitoring data research, particularly in the analysis and prediction of accumulation landslides within reservoir areas.

Debris flows resulting from extreme precipitation exhibit extensive scale, severe erosion along their paths, and substantial sediment deposition, necessitating the implementation of cascade check dam projects in small watersheds for hazard mitigation. In the Wenchuan earthquake-affected area, we investigated 33 typical debris flow gullies and obtained characteristic parameters for 105 check dams, along with the erosion pattern changes in gullies under the influence of cascade check dams. We summarized the types of check dams and their identification criteria and proposed four combinations of solid dams and open dams, as well as two back-silting modes under different spacings of cascade dams. We further analyzed the effects of characteristic patterns of cascade dam (dam type combination mode and back-silting mode) on channel’s erosion and deposition pattern (slope reduction coefficient due to sedimentation and relative erosion depth coefficient). The results indicated that the combination of solid and permeable dams (SO mode and OS mode) significantly influenced the back-silting pattern of channel, and the interactive back-silting mode had a notable impact on deposition pattern, while the independent back-silting operation mode yielded stronger erosion effect. The research findings offer reference for the selection of check dams and the optimization of dam spacing of cascade dams in small watersheds.