In order to investigate the influence of the ion exchange process on the permeability of rare earth ore bodies in the leaching process, a laboratory-scale comparative experiment with ammonium sulfate solution and deionized (DI) water as leaching solutions is conducted. Compared with the DI water leaching test, the permeability coefficient of the rare earth ore sample leached by the ammonium sulfate solution gradually decreases at the beginning and then increases with the completion of leaching. The physical and morphological evolutions of rare earth ore samples in this comparative experiment are also monitored by nuclear magnetic resonance and scanning electron microscopy. It is concluded that the change in the permeability coefficient arises from the adsorption–desorption of a large number of clay microparticles, resulting in a dynamic evolution of pore structures. Further mechanism analysis suggests that the change in internal ionic strength caused by ion exchange and leaching solution seepage promotes the adsorption–desorption behavior of clay microparticles.

Two coupling processes—solution seepage and chemical replacement—occur in the in situ leaching process of ion-absorbed-rare-earth ore. In this study, saturated leaching tests were applied to investigate the permeability characteristics of rare earth ore under the hydro-chemical coupling effect. The pore radius distributions based on two leaching solutions (H₂O and (NH₄)₂SO₄) were obtained by nuclear magnetic resonance detection technology. The results indicated that 10–25 μm and 4–10 μm pores are dominant in the ore under solution leaching using H₂O and (NH₄)₂SO₄, respectively. A “black belt” in the pores of (NH₄)₂SO₄ leaching was discovered from the reconstruction of the pore structure distribution image. The results also reveal that the hydraulic conductivity will initially increase, then decrease and finally increase during the leaching process. The hydro-chemical coupling effect will lead to variations in the pore structure and permeability of rare earth ore. The pore radius will increase due to solution seepage, whereas it decreases due to the particle recombination induced by chemical replacement. The permeability of rare earth ore is influenced more by chemical replacement than by solution seepage.

To investigate the effect of dimensional changes on the mechanical properties of cemented tailings backfill (CTB), uniaxial compression tests are performed on square CTBs of four different sizes. Combining digital image correlation (DIC) and acoustic emission (AE) methods, the fracture process and AE behavior characteristics of backfills with different sizes are analyzed. The results show that as the backfill size increases, its uniaxial compressive strength decreases, and its strength stabilizes gradually when it measures 100 mm. Under uniaxial compression loading, surface cracks on the smaller specimens evolve rapidly and aggressively, with no significant shedding area, whereas the larger specimens show plastic failure. The cracks expand and penetrate gradually, forming a large shedding area. As the specimen size increases, the backwards trend of the peak value of the ringing count relative to the peak value of the stress becomes increasingly evident. Combining the change law between the r value and the average frequency centroid, the sudden drop point of r value and the lowest value of average frequency centroid can be regarded as the precursor of macroscopic damage.

The strength of cemented paste backfill (CPB) directly affects mining safety and progress. At present, in-situ backfill strength is obtained by conducting uniaxial compression tests on backfill core samples. At the same time, it is time-consuming, and the integrity of samples cannot be guaranteed. Therefore guided wave technique as a nondestructive inspection method is proposed for the strength development monitoring of cemented paste backfill. In this paper, the acoustic parameters of guided wave propagation in the different cement-tailings ratios (1:4, 1:8) and different curing times (within 42 d) of CPBs were measured. Combined with the uniaxial compression strength of CPB, relationships between CPB strength and the guided wave acoustic parameters were established. Results indicate that with the increase of backfill curing time, the guided wave velocity decreases sharply at first; on the contrary, attenuation of guided waves increases dramatically. Finally, both velocity and attenuation tend to be stable. When the CPB strength increases with curing time, guided wave velocity shows an exponentially decreasing trend, while the guided wave attenuation shows an exponentially increasing trend with the increase of the CPB strength. Based on the relationship curves between CPB strength and guided wave velocity and attenuation, the guided wave technique in monitoring the strength development of CPB proves feasible.

In order to study the influence of the cement–tailing ratio and curing age on the mechanical properties of cemented backfill with unclassified tailings and improve the accuracy of the backfill monitoring method, the mechanical characteristics of the failure process of cemented backfill with unclassified tailings at different curing ages were studied using the acoustic emission and infrasound method. The results show that the peak strength and modulus of elasticity of backfill are positively correlated with the curing age. During the failure process of cemented backfill with unclassified tailings, the acoustic emission ringing count shows a “stabilizing–rising–falling–rising” trend, and the infrasonic ringing count will have a “multiple peak” phenomenon. The ib value of the acoustic emission signal generally increases with the extension of the curing age. The ib value of the infrasonic signal generally has a negative correlation with the increase of the curing age. When the ib value of the acoustic emission and infrasonic wave all start to decline in a jumping manner, this indicates that the backfill is about to be damaged. The dominant frequency ratio of acoustic emission signals (F_AE) is distributed between 0–4 during the failure process, and the total number of F_AE shows the phenomenon of “first decreasing and then increasing” with the curing age. The dominant frequency ratio of infrasonic signals (F_s) is distributed between 0–6. During the failure process of cemented backfill with unclassified tailings, the quantity ratio of class A and B of acoustic emission signals shows a gradually increasing trend, and the quantity ratio of class A and B of infrasonic signals shows a “decreasing–increasing–decreasing” trend, and the quantity ratio of class A and B of acoustic signals shows a sharp increase at the stage of entering the crack’s unstable growth stage.

Water inrush and mud burst disasters pose severe challenges to the safe and efficient construction of underground engineering. Water inflow prediction is closely related to drainage design, disaster prevention and control, and the safety of the surrounding ecological environment. Thus, assessing the water inflow accurately is of importance. This study proposes a Bayesian Optimization-Long Short-Term Memory (BOA-LSTM) recurrent neural network for predicting tunnel water inflow. The model is based on four input parameters, namely tunnel depth (H), groundwater level (h), rock quality designation (RQD), and water-richness (W), with water inflow (WI) as the single-output variable. The model first processes and analyzes the data, quantitatively characterizing the correlations between input parameters. The tunnel water inflow is predicted using the long short-term memory (LSTM) recurrent neural network, and the Bayesian optimization algorithm (BOA) is employed to select the hyperparameters of the LSTM, primarily including the number of hidden layer units, initial learning rate, and L2 regularization coefficient. The modeling process incorporates a five-fold cross-validation strategy for dataset partitioning, which effectively mitigates overfitting risks and enhances the model’s generalization capability. After a comprehensive comparison among a series of machine learning models, including a long short-term memory recurrent neural network (LSTM), random forest (RF), back propagation neural network (BP), extreme learning machine (ELM), radial basis function neural network (RBFNN), least squares support vector machine (LIBSVM), and convolutional neural network (CNN), BOA-LSTM performed excellently. The proposed BOA-LSTM model substantially surpasses the standard LSTM and other comparative models in tunnel water inflow prediction, demonstrating superior performance in both accuracy and generalization. Hence, it provides a reference basis for tunnel engineering water inflow prediction.