To investigate the influences of endwall roughness and ore grading condition on the ore drawing effect of non-pillar sublevel caving method, based on a single segment and single access ore drawing model, the joint roughness coefficient (JRC) of rock structure surface was introduced to quantify the endwall roughness. Physical experiments were conducted under five types of endwall roughness and two types of ore grading conditions, and the changes in recovery rate, dilution rate, and release amount of ore, as well as the development characteristics of the release body morphology, were systematically analyzed. The results indicate that the endwall roughness and the ore grading conditions have a significant impact on the flowability of the ore, thereby determining the variation characteristics of the recovery rate and dilution rate. As the proportion of fine particles in the ore grading and the endwall roughness decrease, the ore recovery rate shows an upward trend. The dilution rate shows a trend of first increasing, then decreasing, and then increasing again as both factors increase. When the content of fine particles is low and the JRC value is 0, the ore flowability is the optimal, the recovery rate is the highest at 75.15％, and the dilution rate is the lowest at 32.09％. In addition, reducing the proportion of fine particles in the ore grading and decreasing the endwall roughness can promote the development of the transverse axis length and suppress its height development of the ore release body, thereby increasing the ore recovery rate and optimizing the overall ore drawing indicators.

Multi-outlet ore drawing method is commonly adopted in underground mine production, and there is often an obvious arching phenomenon during the ore drawing process. To investigate the influence of multi-outlet ore drawing on the arching of granular ores, based on the theory of lateral pressure of granular materials, a lateral pressure testing system for granular material flow was used to conduct physical similarity simulation experiments under single-outlet, double-outlet, and triple-outlet ore drawing conditions. The distribution laws of lateral pressure of granular ores under different ore drawing methods and filling heights were obtained, so as to further analyze the ore arching phenomenon. The research results show that when the filling height is 75 cm, the ore arching phenomenon under triple-outlet ore drawing is relatively obvious, and the arching strength is the highest. The location of arching is less affected by the ore drawing method, and obvious arching phenomena mostly occur at 55 cm above the ore outlet. Under the condition of triple-outlet ore drawing, the filling height further affects the formation of arching and the location of its occurrence. When the filling height is 85 cm, obvious arching phenomena occur at 55 cm and 75 cm above the ore outlet, and the ore in the lower area shows a tendency for arching. When the filling height increases to 95 cm, the area where obvious arching phenomena occur extends downward.

To further investigate the flow characteristics of the ore drawing process of ore-rock with different water contents in steeply inclined extremely thin veins, so as to optimize mining processes and improve resource recovery rates, a combined method of physical tests and numerical simulation was adopted, and a full-scale numerical model of ore drawing in confined space for steeply inclined extremely thin veins was established. Through repose angle calibration tests on ore-rock with different water contents, the discrete element simulation parameters were determined. Full-scale numerical simulations of ore drawing were carried out, and the ore drawing characteristics of ore-rock with different water contents under different ore drawing heights were analyzed. The results show that the water content of ore-rock particles has a significant impact on the flow characteristics of ore-rock. When the water content of ore-rock is 4％, the ore drawing fluidity is optimal with a large and stable flow rate, and it exhibits a “first-in first-out” flow regime. As the water content increases, the ore drawing fluidity distorts and gradually transitions to a “first-in last-out” flow regime, while the ore drawing flow rate decreases, failing to meet the requirements for simultaneous operation of multiple ore drawing outlets. In addition, when the water content of ore-rock is 4％, ore drawing heights have a relatively small impact on the fluidity of ore-rock particles. As the water content increases, the increase in ore drawing heights worsens the fluidity of particles at higher positions, and when the water content is 15％, the particle fluidity deteriorates severely.

The difficulty of controlling the blasting effect of fan-shaped blasting holes in the non-pillar sublevel caving method significantly affects the mining efficiency in the mining area. Research was conducted on the issues of high large block rate and poor ore recovery index as well as poor safety of mining operations in the 1 580 m sublevel caving stope under the condition of broken ore-body in the west No.2 mining area in Longshou Mine. The analysis shows that the main reason of the high block rate is that the side holes in upper sublevel edge are buried by the blasting pile and not fully blasted, and the serious blockage of the blasting row holes in this section results in insufficient charging. Analysis shows that a high block rate can have adverse effects on the mining indicators, operational safety, mining efficiency, and mining costs of the mining area. In order to reduce the large block rate in the mining area, a series of technical and management measures were proposed, including optimizing the form and size of the mining drift, improving the parameters of fan-shaped blast holes, introducing through hole equipment suitable for the working environment, testing pre-charge technology, and strengthening blasting quality management. The large block rate in the mining area has been successfully reduced from 16％ to 5％, the ore recovery index in the mining area has been increased from 68％ to 82％, and the safety of the mining area has been significantly improved.

Since the filling mining technology was adopted in Jinchuan Longshou Mine, the filling quality has not been systematically studied. Based on this, a follow-up survey of the filling quality of the downward layered cemented filling method in Jinchuan Longshou Mine was carried out for half a year, and the key factors affecting the filling quality were analyzed. Combined with the results of investigation and research, the filling quality control measures were put forward. The survey results show that the quality of mine filling materials is poor, the filling body is not completely connected to the roof, the strength of the filling body fluctuates greatly, and the number of abnormal parking times has a great impact on the filling quality. The field practice shows that the filling quality can be improved by ensuring the unit consumption of cementitious materials and the quality of filling materials, improving the roof-contacted filling rate, reducing the frequency of abnormal parking, improving the uniformity of slurry and ensuring reasonable mass concentration of slurry.

Aiming at the problems of short pumping distance and easy plugging of gravel filling slurry in Datun Tin Mine, it was proposed to add copper slag tailings fine material to the gravel to improve the aggregate gradation, and the gravel-copper slag tailings mixture ratio test was carried out. It was concluded that when the proportion of gravel mass reached 60％, the packing density of mixed aggregate reached the maximum value. Further, through the strength test of backfill, rheological test, high concentration and high flow state test, the mix ratio of gravel-copper slag tailings paste was determined, which meets the strength requirement of filling body, no segregation of paste and self-leveling. Based on the actual filling situation of 32^# stope in Datun Tin Mine, the numerical simulation method was used to analyze the change law of pipeline resistance of gravel-copper slag tailings paste. The results show that the optimal mix ratio of gravel-copper slag tailings paste is gravel∶ copper slag tailings ＝ 6∶4, and the cement dosage are 270 kg/m³ (mass concentration 82％－83％), 240 kg/m³ (mass concentration 83％－84％), 210 kg/m³ (mass concentration 83％－84％), 180 kg/m³ (mass concentration 83％－85％), respectively. The resistance loss of gravel-copper slag tailings paste along the horizontal pipeline is 6 105.76－6 525.11 Pa/m, and the limit pumping horizontal distance is 2 298.81－2 456.70 m. The paste was used for industrial filling with a mix proportion of gravel∶ copper slag tailings ＝ 6∶4, a cement dosage of 240 kg/m³, and a mass concentration of 83％. Compared with the original filling process with multi-stage pumping, the paste with this mix ratio can meet the filling demand only under the condition of single pump, and the single pump conveying distance (2 052 m) is less than the limit pumping horizontal distance, allowing for surplus power at the outlet, which verifies the effectiveness of the optimal mix ratio.

Aiming at the problem of high filling cost and poor process stability caused by insufficient tailings production in a mine, the mechanical and rheological properties of Gobi sand-tailings synergistic filling body were studied. The effects of slurry mass concentration (74％, 76％, 78％), cement-sand ratio (1∶6, 1∶8, 1∶10, 1∶12), Gobi sand content (0, 10％, 20％, 30％) and curing age (3 d, 7 d, 14 d, 28 d) on the compressive strength of filling body and the transport resistance of filling slurry were discussed. Based on the MATLAB platform, the Random Forest (RF) and Convolutional Neural Network (CNN) models were constructed to predict the compressive strength of the filling body. The results show that the uniaxial compressive strength of filling body increases with the increase of slurry mass concentration, curing age and cement-sand ratio, and increases first and then tends to be stable with the increase of Gobi sand content. The transport resistance of filling slurry increases with the increase of slurry mass concentration, and decreases first and then increases with the increase of Gobi sand content. According to the uniaxial compressive strength of the filling body and the slurry transport resistance, the optimal content of Gobi sand is 20％. In the neural network model, the prediction accuracy of the CNN model is better than that of the RF model, determination coefficient R²＝0.970 34, the average absolute error MAE＝0.101 81, and the error between the predicted value and the experimental value is 0－10.53％. The research results can provide reference for the collaborative utilization of low-grade aggregate and the reduction of filling cost in the same type of mines.

After the 1^# orebody of Kalatongke Copper-Nickel Mine was mined by the caving method, a large collapsed loose area was formed. This collapased loose area has deteriorated the mining environment of the upper ore and rock, so that the resources are difficult to be effectively recovered, and the safety of vertical shafts faces a threat. To conduct in-depth detection and treatment of this area, the loose boundary was initially determined through geological drilling and TRT advanced detection in the early stage. Due to channel blockage, traditional detection equipment could hardly access the target area. The C-ALS Gyro endoscopic three-dimensional scanning system was used to conduct detailed detection of the potential area, which clarified the location, shape and volume of the large loose area. It was found that this area only exists locally, and the overall collapse scope tends to be stable. By comparing three treatment schemes including waste rock backfilling treatment, caving surrounding rock treatment and cemented filling treatment, a comprehensive analysis of the advantages and disadvantages of each scheme was conducted. In accordance with the treatment principles, it was determined to adopt the cemented filling treatment scheme. The actual treatment effect shows that the cemented filling treatment scheme enhances regional stability, increases the resource recovery rate by approximately 30％, improves the safety guarantee level, reduces environmental impact.

In order to quantify the effects of different protective measures on the stability of jointed rock slopes, the Particle Flow Code in two-dimensions (PFC2D), strength reduction method and failure criterion characterized by a coherent fracture plane were used to reveal the failure characteristics of jointed rock slope when the parameters of contact model have been reduced until the critical rock failure state had occurred. Then the safety factor and microcrack evolution as well as energy dissipation law of jointed slopes under three working conditions including slope-cutting, anchor cable reinforcement and without any protective measures were comprehensively researched. Research shows that when the mechanical parameters have been reduced until the critical rock failure state has occurred, the main failure mode of jointed rock slope is tensile-sliding failure. The slope-cutting measure can partially improve the slope stability, and the anchor cables installed at a zone where is below the joint surface and close to the the intersection of the steeply inclined fracture and the gentle slip surface can significantly improve the slope stability. During the adjustment process to the stable state, the number of microcracks and dissipated energy generated in anchor cable-reinforced slopes are sequentially less than those in slope-cutting slopes and slopes without any protective measures. The research results can provide a reference for the slope engineering in similar open-pit mines.

To enhance the accuracy and reliability of measuring the thickness of the surrounding rock loosened zone in roadways, a high-precision prediction model (PSO-Stacking) for the thickness of the loosened zone was proposed. This model employed interpretable artificial intelligence algorithms to address the black-box effect of machine learning algorithms and explained the prediction mechanism. The model was validated through an engineering case study at Xinglongzhuang Coal Mine. By incorporating SHAP analysis to interpret model decisions, the influence of surrounding rock loosened zone characteristics was revealed, thereby improving interpretability. The research results demonstrate that the PSO-Stacking model performs exceptionally well in predicting the thickness of the loosened zone, with high prediction accuracy and minimal error. On the test set, the R² value reached 0.957, while the RMSE and MAE were 0.118 and 0.087, respectively. Global interpretability analysis of the prediction model indicates that roadway burial depth and joint development degree have the most significant impact on the prediction results. Local interpretability analysis reveals that burial depth considerably influences the thickness of the loosened zone in high-strength surrounding rock, while joint development has a lesser effect. According to the application in Xinglongzhuang Coal Mine, the average error between the predicted results and the actual values is only 4.31％, demonstrating that the proposed model can effectively predict the thickness of the surrounding rock loosened zone and reveal the influence of its characteristics. This study provides timely scientific reference for evaluating surrounding rock support and its effectiveness, meeting the demands of engineering practice.