In order to avoid the phenomenon that the good rock is over-supported and the poor rock is under-supported, based on the XGBoost algorithm combined with the roadway safety factor, an intelligent evaluation and optimization design system of on-demand support in Longqiao Iron Mine was developed. 7 on-demand anchor mesh support indexes and 5 on-demand plain spray support indexes affecting the stability of roadway in Longqiao Iron Mine were used as the main input values. 60 sets of practical roadway data were selected as examples to carry out the on-demand optimization support design and effectiveness evaluation. The results show that at the level of −370.0 m, the surrounding rock grade is better, the safety factor is larger, and the support measures are not needed. At the level of −382.5 m and −395.0 m, the surrounding rock grade is slightly worse, the safety factor is slightly lower, and the plain spray support form is required. At the level of −407.5 m and −420.0 m, the surrounding rock grade is above the third grade, and the surrounding rock properties are poor, so it is necessary to adopt the form of anchor mesh support. The intelligent evaluation and optimization design system of on-demand support provides real-time rock mass quality evaluation and support scheme optimization for Longqiao Iron Mine, which has effectively improved the safety of roadway and significantly saved the support cost.

Rainfall is a primary factor inducing landslides on slopes, especially the slope repaired after in-situ leaching of ionic rare earth mine is more susceptible to rainfall infiltration. Taking the slope of an ionic rare earth mine in South China as the research background, the influences of rainfall, duration and rainfall type on slope stability was systematically analyzed based on the monitoring data of soil moisture and finite element simulation. The results show that the proportion of deep infiltration (deeper than 60 cm) events (accounting for 19.79%) of single-peak and multi-peak rainfall is significantly higher than that of concentrated (accounting for 1.04%) and uniform (accounting for 5.21%). The numerical simulation shows that the slope stability under the condition of low rainfall (10−20 mm/d) is concentrated>uniform>single peak>multi peak. The stability of single peak type is the worst when the rainfall is high (50 mm/d). With the increase of rainfall duration, the slope stability coefficient under concentrated and uniform rainfall conditions shows a trend of slow decline and then rapid decline, and the slope stability coefficient under single-peak and multi-peak rainfall conditions shows a trend of rapid decline and then slow decline, which is closely related to the sudden change of matrix suction caused by peak rainfall intensity. The research can provide a theoretical basis for the prevention and control of rainfall-induced landslides in ionic rare earth mines.

In order to study the influence of prefabricated fracture length on the tensile properties and fracture deformation characteristics of different rock materials, notched semi-circular bend (NSCB) specimens of three typical brittle rocks, including granite, medium sandstone and limestone, were selected for loading tests. The deformation of the specimen during failure was monitored by digital image correlation (DIC). The results show that the fracture toughness of rock specimens is positively correlated with mechanical parameters such as compressive strength, tensile strength and elastic modulus, and the bearing capacity of NSCB specimen is negatively correlated with the length of existing cracks. When the fracture length exceeds a certain value(a≥3 mm), the bearing capacity of NSCB specimen has a negative exponential relationship with the fracture length. The deformation characteristics of NSCB specimens are significantly affected by the crystal structure and micro-crack distribution of rock. The higher the degree of cementation of rock minerals, the denser the rock crystal structure, the less developed the micro-cracks, the higher the fracture initiation stress, and the smaller the deformation degree before failure, and vice versa. The research results can provide a basis for further understanding the fracture propagation mechanism of rock materials and predicting its expansion range.

In the process of mineral exploitation, rock mass often bears different degrees of dynamic load. A series of dynamic impact and uniaxial compression tests were conducted to analyze the characteristic stress and acoustic emission characteristic parameters of sandstone after impact. The influence of dynamic disturbance damage on the mechanical behavior and acoustic emission response characteristics of sandstone specimens was studied. The results show that with the increase of impact velocity, the longitudinal wave velocity of sandstone specimens after impact test decreases gradually, while the internal damage variable increases gradually. With the increase of impact velocity, the ratio of crack closure stress to peak stress of sandstone specimens increases gradually, while the ratio of crack initiation stress to peak stress and the ratio of damage stress to peak stress decrease gradually. In the process of rock deformation and failure, acoustic emission activity shows significant stage evolution characteristics, and its activity intensity and frequency characteristics will change significantly with the damage evolution inside the rock. The research results can provide theoretical basis for the stability evaluation and support design of roadway surrounding rock under blasting disturbance.

In order to realize the accurate prediction of mine dust concentration, based on the field monitoring data of a mining working face in a coal mine in Shaanxi Province, the random forest (RF) model was used to construct a multi-factor dust concentration prediction model by taking temperature, humidity, wind speed and dust concentration at the current moment as input vectors. Four intelligent optimization algorithms (Newton-Raphson-based optimizer (NRBO), sparrow search algorithm, grey wolf optimizer and whale optimization algorithm) were proposed to optimize the parameters of random forest (number of trees and number of leaf nodes). The model was evaluated by determination coefficient(R²), root mean square error (RMSE) and mean absolute error (MAE). The results show that the optimal model for dust concentration prediction is NRBO-RF model. Compared with RF model, the R², RMSE and MAE of NRBO-RF model are increased by 0.030 3, 0.572 8 and 0.074 9, respectively, and its fitting ability and prediction accuracy are the highest. The research results can provide a reference for intelligent prediction and prevention and control of mine dust.

To explore the impact mechanism of ecological restoration project on the migration and release of antimony in groundwater in TinMine of Hunan, hydrochemical comparisons and hydrogeochemical simulations were conducted on groundwater samples collected from TinMine in 2016 and 2023. The changes in groundwater hydrochemical characteristics and antimony content before and after ecological restoration were analyzed. The results show that ecological restoration has varying degrees of impact on the antimony content in three different limestone fracture aquifers in the mining area, among which the Shetianqiao formation and TinMine formation are significantly affected by mining activities and external pollution, and the antimony mass concentration generally increases. The level of groundwater in the lower carboniferous formation aquifer is low and there is an aquitard, resulting in a decrease in antimony content. The source of Sb before and after ecological restoration has strong differentiation, and its migration and transformation are significantly affected by large-scale soil and green covering projects. The ecological restoration project has changed the redox environment and accelerated the oxidation of Sb (Ⅲ) in groundwater. The main form of antimony in groundwater is . The research results can provide scientific basis for the prevention and control of heavy metal pollution in ecological restoration projects in mining areas.

In view of the scarcity of topsoil and the shortage of solid waste resources in the northwest mining area, the soil reconstruction technology based on coal gangue in the mining area was studied. The coal gangue (CG), rock and soil stripping (RS) and topsoil (TS) from a coal mine in Tongchuan, Shaanxi Province were selected as raw materials, and the reconstructed soil was mixed in different proportions to carry out the pot experiment of alfalfa. The results show that the physical and chemical properties of the reconstructed soil under different proportions were different. In general, the reconstructed soil increases the nutrient and moisture content of the soil, which is more conducive to the growth and development of alfalfa. Compared with TS, the pH of the reconstructed soil decreases, the organic matter, total nitrogen and available potassium increase, while the available phosphorus decreases. The treatment of CG 40% and RS 15% is the most favorable for the growth of alfalfa. The highest plant height of alfalfa is 54.60 cm, the longest root length is 23.20 cm, and the maximum fresh weight and dry weight are 14.814 g and 4.849 g, respectively. The above results indicate that the reconstructed soil formed by CG is feasible for the phytoremediation of mines in Northwest China, and provides a new choice for the comprehensive utilization of CG.

In order to strengthen the safety management of non-coal underground mines and clarify the characteristics of hidden dangers in non-coal underground mines, the total number of hidden dangers, types of hidden dangers and characteristics of major hidden dangers in 52 non-coal underground mines were statistically analyzed. The results show that the number of general hidden dangers is about 11.8 times that of major hidden dangers, and the number of on-site management hidden dangers is about 1.1 times that of basic management hidden dangers. The number of major hidden dangers of on-site management is about 1.3 times that of basic management. Major hidden dangers are mainly concentrated in Article 23, Article 20, Article 4 and Article 31, which are 28, 21, 19 and 13 items respectively. In view of the characteristics of major hidden dangers, it is proposed that mining enterprises should strengthen the construction of organizations, fully invest in safety funds and improve and improve safety facilities, as well as safety management departments should extend supervision tentacles, innovate supervision modes and strengthen supervision compatibility. The research results will provide reference for the safety management of non-coal underground mines.

Aiming at the problem of difficult diagnosis of mining truck drive system, an intelligent fault diagnosis model was proposed by combining wavelet packet and ensemble empirical mode decomposition (EEMD) method, which was integrated into intelligent operation and maintenance system and applied to mine truck. Firstly, the wavelet packet was used to denoise the vibration signals of bearings and gears in the drive system. Then the EEMD algorithm was used to decompose the signal into several intrinsic mode function (IMF) components, and the feature vector was extracted from the IMF components containing the main state information. Secondly, the support vector machine (SVM) was used to classify the fault modes. The penalty parameter C and kernel function parameter h of SVM were optimized by particle swarm optimization (PSO), genetic algorithm (GA) and multi-population genetic algorithm (MPGA). Four fault diagnosis models of SVM, PSO-SVM, GA-SVM and MPGA-SVM were constructed respectively. It is concluded that MPGA-SVM is the optimal fault diagnosis model, and the diagnostic accuracy of bearing fault and gear fault is the highest, which is 98.75% and 6.67% respectively. Finally, aiming at the RE1060 drive system of XDE320 electric-driven mining dump truck, an intelligent operation and maintenance system including four modules of signal acquisition and storage, time-frequency analysis, fault feature extraction and intelligent fault diagnosis was developed by MATLAB and LabVIEW. The research shows that the intelligent operation and maintenance system can realize on-line monitoring and diagnosis of existing faults, which is of great significance to improve the safe and efficient operation of mining trucks.

In order to study the safety design and intelligent design of multi-stage pipeline oil vertical transportation system in underground mines, the numerical relationship model between maximum thrust and single rod length in the installation process and the dynamic vibration analysis model of vertical oil pipeline with different section sizes and different lengths were established. The numerical model of water hammer effect and pressure relief transportation of oil transportation system was built, and the dynamic response characteristics of water hammer pressure under different oil flow and different valve closing time conditions were studied and analyzed. The results are as follows. Firstly, when the length of a single pipe is at a certain value, the installation thrust increases sharply with the increase of pipe diameter. Secondly, the pipe sand filling method can reduce the vibration of the main pipe caused by bubble excitation. Thirdly, reducing the oil flow or increasing the valve closing time can effectively reduce the water hammer pressure. At the same time, the intelligent design schemes of intelligent refueling technology, intelligent dispatching technology, intelligent monitoring technology, intelligent fire protection and fire automatic extinguishing technology are proposed, which plays an important role in the in-depth study of the safety intelligent design of pipeline oil vertical transportation system.