Hongyi Zhang

The primary objective of this research is to quantitatively isolate the complex driving factors of slope deformation and explicitly reveal the long-term creep mechanism induced by early excavation unloading, thereby providing a theoretical basis for long-term stability evaluation. To achieve this, this study adopts a combined approach of multivariate statistical regression and numerical simulation inversion based on long-sequence monitoring data. First, a multivariate statistical regression model incorporating time-dependent, rainfall, temperature, valley width, and excavation components was constructed to quantitatively separate the contribution weights of each factor. Second, by introducing a rock–soil creep constitutive model, a refined finite element model was established to perform back-analysis of creep parameters and numerical simulation. The results indicate that two large-scale slope-cutting excavations were the direct triggers for the deformation, resulting in shear dislocation of the deep ancient sliding zone and superficial slippage. The dominant factors exhibit distinct phasic and spatial differences: before impoundment, the time-dependent component was absolutely dominant (>80%); after impoundment, low-elevation areas were significantly affected by valley width shrinkage (>60%), while high-elevation areas remained dominated by time-dependent deformation (>74%). Numerical simulation confirmed that the nature of the deformation is “excavation unloading-induced creep along the ancient sliding zone,” and the simulation results considering creep effects accurately reproduced the actual deformation characteristics observed in situ. It is concluded that the rheological effects induced by early excavation unloading are central to the control of long-term stability.

Difficulty in clarifying the deformation characteristics of deep rocks under a high water pressure environment is a technical bottleneck restricting the safe operation of large hydropower stations. In order to study the effect of reservoir water level changes on the mechanical behavior of deep limestone, a series of mechanical tests were conducted under different dynamic high water pressure environments using a self-developed hydraulic loading test device. The test results show that the unsaturated limestone always undergoes compressive deformation during the linear increase in external water pressure, and the saturated limestone changes its deformation state from compression to expansion during the linear decrease in external water pressure. The stress–strain curve of limestone shows apparent hysteresis characteristics during the cyclic increase and decrease in external water pressure. Overall, the rock strain rate showed a significant negative correlation with the external water pressure, and the rock deformation modulus showed a certain positive correlation with the external water pressure. During hydraulic loading, saturated rocks had a smaller range of variation in the strain rate and deformation modulus and were more resistant to deformation than unsaturated rocks. Limestone was subjected to both external water pressure and internal pore water pressure in a cyclic cycle, where pore water pressure promotes pore creation and expansion, while external water pressure prevents water from degrading the pore structure. The periodic change of water pressure has a significant influence on rock mechanics and deformation behavior, and the rock mass will undergo elastic deformation, plastic deformation, and even fracture. Further study of this deformation rule can provide a more accurate theoretical basis for the safe operation of water conservancy projects.

In response to the challenge of limited model availability for predicting the lifespan of super-high arch dams, a hybrid model named EMD-PSO-GPR (EPR) is proposed in this study. The EPR model leverages Empirical Mode Decomposition (EMD), Gaussian Process Regression (GPR), and Particle Swarm Optimization (PSO) to provide an effective solution for super-high arch dam stress prediction. This research focuses on three strategically selected measurement points within the dam, characterized by complex stress conditions. The predicted results from the EPR are compared with those from GPR, Long Short-Term Memory (LSTM), and Support Vector Regression (SVR), using actual stress data measured at research points within a super-high arch dam in Southwest China. The findings reveal that the proposed EPR model attains a maximum mean absolute error (MAE) of 0.02916 and a maximum root mean square error (RMSE) of 0.03055, surpassing the compared models. As a result, the EPR model introduces an innovative computational framework for stress prediction in super-high arch dams, excelling in handling stress data characterized by high vibration frequencies and providing more accurate predictions.