The initial in situ stress field influences underground engineering design and construction. Since the limited measured data, it is necessary to obtain an optimized stress field. Although the present stress field can be obtained by valley evolution simulation, the accuracy of the ancient stress field has a remarkable influence. This paper proposed a method using the generative adversarial network (GAN) to obtain optimized lateral stress coefficients of the ancient stress field. A numerical model with flat ancient terrain surfaces is established. Utilizing the nonlinear relationship between measured stress components and present burial depth, lateral stress coefficients of ancient times are estimated to obtain the approximate ancient stress field. Uniform designed numerical tests are carried out to simulate the valley evolution by excavation. Coordinates, present burial depth, present lateral stress coefficients and ancient regression factors of lateral stress coefficients are input to GAN as real samples for training, and optimized ancient regression factors can be predicted. The present stress field is obtained by excavating strata layers. Numerical results show the magnitude and distribution law of the present stress field match well with measured points, thus the proposed method for the stress field inversion is effective.

As new ways to solve partial differential equations (PDEs), physics-informed neural network (PINN) algorithms have received widespread attention and have been applied in many fields of study. However, the standard PINN framework lacks sufficient seepage head data, and the method is difficult to apply effectively in seepage analysis with complex boundary conditions. In addition, the differential type Neumann boundary makes the solution more difficult. This study proposed an improved prediction method based on a PINN with the aim of calculating PDEs with complex boundary conditions such as Neumann boundary conditions, in which the spatial distribution characteristic information is increased by a small amount of measured data and the loss equation is dynamically adjusted by loss weighting coefficients. The measured data are converted into a quadratic regular term and added to the loss function as feature data to guide the update process for the weight and bias coefficient of each neuron in the neural network. A typical geotechnical problem concerning seepage phreatic line determination in a rectangular dam is analyzed to demonstrate the efficiency of the improved method. Compared with the standard PINN algorithm, due to the addition of measurement data and dynamic loss weighting coefficients, the improved PINN algorithm has better convergence and can handle more complex boundary conditions. The results show that the improved method makes it convenient to predict the phreatic line in seepage analysis for geotechnical engineering projects with measured data.

The in situ stress field is the fundamental factor causing deformation and damage in geotechnical engineering, so it is the main basis for underground engineering design and excavation. However, it is difficult to accurately obtain the in situ stress through most existing inversion methods in areas with complex geological conditions. For the problem of a relatively discrete and nonlinear relationship of measured stress in the Yebatan Hydropower Station area, a new in situ stress inversion method called the local stress field correction (LSFC) method combining a genetic algorithm (GA), backpropagation (BP) neural network, and submodel method is proposed. The inverted in situ stress results produced by this method show that the distribution of in situ stress is greatly influenced by tectonic movements in the Yebatan area, there is no obvious linear relationship with depth, and the stress release phenomenon occurs at the faults. By comparison with the multiple regression method, it is found that the method still has high inversion accuracy under complex geological conditions, and the average relative error of LSFC inversion results is 17.05%, which is much lower than the value of 43.58% via the multiple regression method. Therefore, the LSFC method can be used for the inversion of in situ stress in complex geological regions and provide a reference for engineering design and construction.

Accurate characterization of rock anisotropy is crucial for underground engineering stability assessment. In this study, multi-directional drilling tests were performed on sandy mudstone and argillaceous sandstone, with real-time monitoring of feed force (F), torque (M), rotational speed (n), power (P), drilling velocity (v), and depth (h). Drilling-derived apparent friction angles (φ) in different directions were estimated using a force-equilibrium-based model. Based on drill bit geometry, an energy balance model for hollow drilling was developed, and a unit grinding energy (η_e) was derived. A preliminary drilling-derived anisotropy index ($B_{\phi -{\eta }_e}$), based on the coefficient of variation (CV) and integrating φ and η_e, was proposed. Results show a strong linear correlation between thrust force and torque, both exhibiting a two-stage increase with drilling depth. For the tested drilling orientations, the anisotropy determined using the proposed method decreases in the following order: sandy mudstone 1, sandy mudstone 2, argillaceous sandstone 2, and argillaceous sandstone 1. The proposed index provides a preliminary basis for evaluating directional variations in rock anisotropy.

As a key structural component of rockfill dams, the load-bearing capacity of large-sized concrete slabs under complex multi-axial stresses is directly related to the long-term safe operation of the dams. This study conducted uniaxial and biaxial lateral compression strength tests on C25 concrete slabs with dimensions of 1500 × 1500 × 150 mm using a large-scale bi-directional loading reaction frame test system, systematically revealing the mechanical properties and failure criteria of large-sized concrete slabs. The results indicate that the biaxial compressive strength of the concrete slabs is significantly greater than the uniaxial compressive strength. The stress–strain curves of the concrete slabs and standard specimens exhibit good consistency before failure. Based on uniaxial compressive strength data, the concrete size effect strength reduction formula proposed by Neville was modified, and a compressive strength prediction formula applicable to large-sized concrete members was established. Further integration with code-specified failure criteria led to the development of a biaxial failure envelope for large-sized concrete slabs, which was validated to agree well with measured data. The research findings can provide reliable experimental evidence and theoretical support for the strength reduction, load-bearing capacity assessment, and revisions of relevant design codes for large hydraulic components such as concrete face slabs in rockfill dams.