Based on the practical engineering background of separation of drilling and blasting operations in an iron mine, research was carried out on prominent problems induced by the separation of drilling and blasting operations, including difficulties in production organization and coordination, lagged inter-process information transmission, information silos in blasting operations, frequent operational risks, and high production costs. Using blasting accident case analysis and fault-tree-based risk assessment method for blasting operations, an intelligent management and control scheme for blasting operations in multiple working faces was established. Based on the Vue+Django framework, an intelligent blasting operation management and control system featuring real-time interaction, process task allocation, precise management of blasting materials, and multi-level personnel control was designed and developed. Results show that, under weak-signal conditions, the system maintains an information transmission delay of approximately 250 ms, a packet loss rate of 0.3%, and a reconnection time of less than 3 s. Under a concurrency level of 1 000 users, the system maintains an average response time of less than 1 s and a transaction success rate exceeding 98%. Following field application, the risk frequency of blasting operations is significantly reduced, considerable costs are saved, and blasting efficiency is improved, thereby effectively solving the coordination problems between drilling and blasting operations. The obtained results provide a feasible technical approach for the intelligent upgrading of mine blasting operations.

Point cloud data of mine roadways acquired by three-dimensional laser scanning often contain noise interference. Existing filtering methods struggle to balance point cloud data simplification with the accurate preservation of roadway geometric features. In contrast, adaptive filtering can dynamically adjust filtering parameters based on the characteristics of the original data, thereby providing a high-quality data foundation for subsequent roadway deformation monitoring. An adaptive progressive point cloud filtering algorithm was proposed, which integrated an improved adaptive voxel filter and a layer-constrained dynamic radius filter to achieve precise point cloud denoising. Specifically, the improved voxel filter can adaptively adjust voxel grid sizes according to the geometric complexity of the roadway, effectively simplifying the point cloud while better preserving its geometric features. The layered constraint dynamic radius filter optimized the dynamic radius parameters through a layered constraint model, enabling the accurate elimination of residual noise. Experimental results show that, compared with traditional voxel filtering, the improved voxel filter, and their conventional combinations, the proposed algorithm achieves a point cloud simplification rate exceeding 92% while controlling the change rates of point cloud volume and surface area within 3%. It can provide reliable data preprocessing technical support for mine roadway deformation monitoring.

To address the frequent accidents caused by uneven illumination, complex backgrounds, and human rule-breaking operations in underground coal mines, a personnel behavior recognition method based on the fusion and optimization of improved Retinex and YOLOv5 was proposed. Firstly, an improved multi-scale Retinex algorithm was employed to enhance underground images. Through multi-scale fusion and color restoration techniques, the interference of low illumination, coal dust, and water mist on the images was eliminated, thereby improving image clarity and color fidelity. Subsequently, the enhanced image data were input into YOLOv5 network to construct a target detection model based on the fusion and optimization of Retinex and YOLOv5, which was adapted to detection scenarios with dense equipment and complex backgrounds in underground mines. The experimental results show that the average accuracy of this method for recognizing helmet-wearing behavior and on-site operational violation postures of underground personnel reaches over 95% and 85%, respectively. This method can achieve the precise identification and judgment of rule-breaking behavior by underground personnel, providing technical support for intelligent safety supervision in coal mines.

Based on the tailings-free mine concept, beneficiation and comprehensive utilization of tailings resources were studied for a magmatic hydrothermal gold mine in Hunan. A tailings-free mine technical system was established, adopting the process of gold recovery by a gravity-flotation combined process, quartz comprehensive recovery from gold tailings, and full-volume backfill of residual tailings. The technical route is as follows: gold is efficiently recovered via a vertical centrifuge gravity-flotation combined process, and a gold concentrate with an Au grade of 81.69 g/t and a recovery rate of 97.06% is obtained. Quartz is recovered from gold tailings by the desliming-high intensity magnetic separation-reverse flotation process, and a quartz concentrate with an SiO₂ grade of 98.86% and a recovery rate of 65.37% is obtained. The residual tailings with a yield of 38.23% are all applied to underground backfill. The process innovation effectively solves the problem of tailings disposal, realizes full-component resource utilization of ore and zero tailings discharge, and provides a technical reference for the green development of similar mines.

Aiming at the difficulty of foam size extraction in flotation process, a flotation foam size extraction method based on improved YOLOv11n-seg was proposed. Firstly, the Cross-Scale Feature Fusion Module (CCFM) was introduced to optimize the neck structure and enhance the adaptability of the model to multi-scale foam and the ability of small target detection. Secondly, the Mixed Local Channel Attention (MLCA) mechanism was embedded, and the channel information, spatial information and local information were integrated to improve the expression ability of the network. Finally, the Unified-IoU loss function was introduced to enable the model to dynamically shift attention from the low-quality prediction box to the high-quality prediction box, balancing training speed and detection accuracy. The experimental results show that compared with the traditional watershed algorithm and Mask R-CNN based on deep learning, the proposed method has significant advantages in segmentation accuracy and real-time performance. Compared with the benchmark model YOLOv11n-seg, the optimized model improves the accuracy of mAP@ 0.5 and mAP@ 0.5∶0.95 by 1.5% and 5.3%, respectively, the recall rate increases by 5.1%, the amount of computation decreases by 11.8%, and the number of parameters decreases by 32.1%. This method has a faster detection speed and higher detection accuracy, which can meet the needs of accurate extraction of flotation foam size.

To achieve the efficient exploitation of refractory bismuth-bearing polymetallic ores and the comprehensive recovery of bismuth resources, the technical challenges associated with high-carbon and high-calcium bismuth middlings were addressed, including strong interference from high-carbon and high-calcium gangue, a complex flotation pulp environment, and the inhibition of bismuthinite floatability caused by residual reagents from prior separation stages. An innovative combined impurity removal process of "reagent removal pretreatment-flotation bismuth extraction-acid leaching purification" was proposed. Process mineralogy analysis and bismuth recovery experiments were systematically conducted, and the process parameters were systematically optimized. The research results show that after conditioning with hydrochloric acid and reagent removal treatment with activated carbon, the high-carbon and high-calcium bismuth middlings undergo flotation bismuth extraction using sodium silicate and lime as depressants, butyl xanthate as the collector, and No.2 oil as the frother. Subsequently, acid leaching is performed with hydrochloric acid at a mass concentration of 18% under conditions of a leaching temperature of 30 ℃ and a liquid-solid ratio of 2∶1 for 8 min. Ultimately, a qualified bismuth concentrate with a yield of 17.07%, a bismuth grade of 19.91%, and a bismuth recovery of 74.75% can be obtained. This process precisely overcomes the key technical bottleneck in the recovery of high-carbon and high-calcium bismuth middlings. The reagent removal pretreatment with activated carbon accurately eliminates the inhibitory effect of residual reagents. Subsequently, the flotation process achieves efficient separation and enrichment of bismuth minerals from gangue minerals such as calcium carbonate. Finally, acid leaching realizes the selective removal of residual impurities, thereby significantly enhancing the quality of the bismuth concentrate and ultimately achieving efficient impurity removal and purification of these high-carbon and high-calcium bismuth middlings.

In a medium-grade siliceous-calcareous phosphate ore from the Yichang region of Hubei Province, the fluorapatite is the target mineral and potash feldspar, dolomite, and quartz are the main gangue minerals, and the P₂O₅ grade of the raw ore is 23.27%. To enhance its utilization efficiency of the ore, a double reverse flotation process was adopted, and systematic optimization tests were conducted on grinding fineness, dosages of collectors (H-1, KM-6), and auxiliary reagents (phosphoric acid, HF-959). The flotation mechanism was revealed through measurements of contact angle, Zeta potential, and infrared spectroscopy. The results show that under the grinding fineness of 75 μm accounting for 81%, a closed-circuit flotation flowsheet consisting of one roughing and two sweeping for desilication followed by one roughing and one sweeping for demineralization, a high-quality phosphate concentrate with P₂O₅ grade of 32.20% and recovery rate of 81.20% can be obtained. The Al₂O₃ and MgO contents in the concentrate are reduced to 2.02% (a decrease of 53.35%) and 0.62% (a decrease of 60.26%), respectively, effectively achieving the separation of apatite from potash feldspar, dolomite, and other gangue minerals. Research indicates that the collector H-1 primarily adsorbs physically on the surface of potash feldspar, selectively enhancing its hydrophobicity, which is the core mechanism for the efficient removal of potash feldspar.

To elucidate the electrochemical mechanism of the interaction between butyl xanthate (KBX) and gold minerals under alkaline conditions, and to provide theoretical support for the optimization of alkaline flotation processes for gold-bearing sulfide ores, unidirectional voltammetry, Tafel polarization curve method, and galvanostatic step technique were employed to systematically investigate the interfacial electrochemical behavior of a gold electrode in butyl xanthate solution at pH=9.18. The results indicate that the oxidation reaction of KBX on the gold electrode exhibits mixed electrochemical-mass transfer control characteristics. The oxidation product is the dimer X₂, and the forward reaction rate constant is 1.524 cm/s. The interaction between the gold electrode and the KBX system intensifies with increasing collector concentration. The chemisorption of butyl xanthate on the gold surface constitutes the critical rate-controlling step of the reaction. Process parameters such as grinding fineness and stirring intensity significantly affect the interaction efficiency between the collector and the mineral. The exchange current density for the reduction on the gold electrode surface, measured by the galvanostatic step technique, is 0.075 A/m², which is in high agreement with the corrosion current density obtained from the Tafel method, thereby verifying the reliability of the results.

As global mining is transforming towards green and sustainable development, open-pit mines are generally confronted with a severe challenge that existing tailings storage facilities encroach on the stope boundary during slope expansion and deepening. To address the problem of "tailings encroachment", a new integrated model for full-cycle tailings management featuring "hydraulic-mechanical collaboration and multi-objective optimization of resources, space and safety" was proposed based on a tailings management project of a copper-cobalt mine in Africa. A theoretical framework integrating "objective coordination, technical coordination and management coordination" was established by this model. Its core lies in adopting the spatiotemporal coupling technology of hydraulic mining with large cutter suction dredgers and mechanical excavation, so as to realize efficient recovery of tailings resources and safe release of stope space simultaneously. An integrated "space-air-ground" monitoring network was deployed in the project, and a proactive control strategy of "phased water impoundment and dynamic early warning" was innovatively adopted. The safety of operations adjacent to the high and steep stope slope is guaranteed by these measures. During construction, the maximum horizontal displacement of the north-side seepage cut-off dam is maintained at 67.4 mm. Practice shows that approximately 8 320 kt of tailings are treated and over 4.2×10⁶ m³ of stope space are released by this model. Distinct economic, safety and environmental benefits are achieved, and a replicable and systematic technical solution is provided for this common problem in the mining industry.

The mining of building material quarries generally suffers from problems such as the use of high-quality resources for low-end applications and the indiscriminate mixing of superior and inferior materials, resulting in inconsistent product quality, diminished value, and severe resource waste. To address this, taking granite as the research object, a multi-dimensional comprehensive evaluation system for the optimal use of granite was constructed. This system comprised four first-level indicators, namely appearance quality/chemical performance, physical and mechanical properties, economic benefits, and ecological benefits, along with 19 second-level indicators. Concurrently, a decision model for the optimal use of granite was established, integrating multi-dimensional factors including physical and chemical properties, economic benefits, and environmental impact. To verify the practicability and reliability of the model, Yunfu granite, Luotian granite, and Golden Ma granite were selected as case studies. The results determined that the optimal uses for Yunfu granite, Luotian granite, and Golden Ma granite are natural stone for plazas (roads or pavements), natural stone slabs for furniture, and pebbles and crushed stone for construction, respectively. Further comparative analyses of the calculation results were conducted using the VIKOR method and the grey correlation degree method, which confirmed the feasibility of the AHP-EWM-TOPSIS decision model. Meanwhile, sensitivity analysis verified that the proposed AHP-EWM-TOPSIS decision model possesses good ranking stability. The research results can provide a scientific basis for the optimal use planning of granite resources. The constructed decision model and the developed decision analysis tools have broad application prospects and can effectively support the optimized allocation of granite products.