Object detection in low-light scenarios has been widely acknowledged as a significant research area in the field of computer vision, presenting a challenging task. Aiming at the low detection accuracy of mainstream single-stage object detection models in low-light scenarios, this paper proposes a detection model called DK_YOLOv5 based on YOLOv5, specifically designed for such scenarios. First, a low-light image enhancement algorithm with better results is selected to generate enhanced images that achieve relatively better visual effects and amplify target features. Second, the SPPF layer is improved to an R-SPPF module with faster inference speed and stronger feature expression ability. Next, we replace the C3 module with the C2f module and incorporate an attention mechanism to develop the C2f_SKA module, enabling richer gradient information flow and reducing the impact of noise features. Finally, the model detection head is replaced with a decoupled head suitable for the object detection task in this scenario to improve model performance. Additionally, we expand the Exdark dataset to include low-light data of underground mine scenario targets, named Mine_Exdark. Experimental results demonstrate that the proposed DK_YOLOv5 model achieves higher detection accuracy than other models in low-light scenarios, with an mAP0.5 of 71.9% on the Mine_Exdark dataset, which is 4.4% higher than that of YOLOv5.

In back-fill mining, how to rapidly increase the concentration of tailings is an important problem facing mining engineers. In this paper, the effects of ultrasound frequency (17-25 kHz), power (50-100 W), duration (5-20 min) and start time (3-12 min) on the final underflow concentration (FUC) of unclassified tailings (UTs) were investigated. The flocculation-sedimentation and thickening of tailings were compared with and without ultrasound application. The response surface method was applied to analyze the primary and secondary relationships and interactive relationships between the various ultrasound operating parameters and the FUC, and the optimal conditions were determined. In addition, Environmental Scanning Electron Microscope (E-SEM) was used to analyze the structural changes of underflow aggregates and clearly demonstrated a denser underflow after ultrasound treatment. The results indicated that ultrasound can significantly improve the underflow concentration of the UTs slurry. The frequency and power are the most important influencing factors. The best conditions for ultrasound application are a frequency of 20.4 kHz, power of 90 W, duration of 6.2 min and start time at 15.0 min. The FUC reached 71.75% after several minutes of sonication, which is 4.31% higher than the FUC of free flocculation. The results of E-SEM analysis showed significant differences among the microstructures of settled tailings (STs) after free flocculation and 17 and 20 kHz ultrasound treatment. Ultrasound at a frequency of 20 kHz has a more effective mechanical vibration and cavitation action and can therefore effectively break the polymer flocculant chains into shorter chains and promote the compaction of tailings and the release of water. The size distributions of the flocs before and after sonication also support this conclusion.

Tailings dams, as essential mining structures, are being built globally for containing the chief waste stream of the mining industry. Catastrophic tailings dam breaches have occurred frequently over the past decade, causing severe impacts on the environment, economy, and human health. The foreknowledge of the tailings dam breach overland flow is crucial for the risk assessment and emergency response planning in order to prevent or minimize possible losses. Using unmanned aerial vehicles (UAV) photogrammetry and smoothed particle hydrodynamics (SPH) numerical method, this study proposed a multidisciplinary procedure for modelling a hypothetical tailings dam breach run out flow over the downstream complex terrain. A case study on a 97-m-height tailings dam in Shandong Province of China was carried out. The proposed procedure was proven applicable to determine the overland tailings flow. The submerged area and flow velocities suggested that the downstream G2 highway would hardly be threatened and more concerns should be paid on the factory plants and workers deployed between the dam toe and the highway. Additionally, the application of UAV photogrammetry in the mining industry as a supplementary surveying method can be further expanded, especially for the numerous small-scale mining sites. The proposed procedure is then recommended for the safety management of the tailings’ storage facilities globally.

Tailings dam failure accidents occur frequently, causing substantial damage and loss of human and animal life. The prediction of run-out tailings slurry routing following dam failures is of great significance for disaster prevention and mitigation. Using satellite remote sensing digital surface model (DSM) data, tailings pond parameters and the advanced meshless smoothed particle hydrodynamics (SPH) method, a 3D real-scale numerical modelling method was adopted to study the run-out tailings slurry routing across real downstream terrains that have and have not been affected by dam failures. Three case studies, including a physical modelling experiment, the 2015 Brazil Fundão tailings dam failure accident and an operating high-risk tailings pond in China, were carried out. The physical modelling experiment and the known consequences were successfully modeled and validated using the SPH method. This and the other experiments showed that the run-out tailings slurry would be tremendously destructive in the early stages of dam failure, and emergency response time would be extremely short if the dam collapses at its full designed capacity. The results could provide evidence for disaster prevention and mitigation engineering, emergency management plan optimization, and the development of more responsible site plans and sustainable site designs. However, improvements such as rheological model selection, terrain data quality, computing efficiency and land surface roughness need to be made for future studies. SPH numerical modelling is a powerful and advanced technique that is recommended for hazard assessment and the sustainable design of tailings dam facilities globally.

Efficient thickening of unclassified tailings slurry (UTS) is critical for enhancing mine backfill efficiency and reducing operational costs. Ultrasonic technology has emerged as a promising approach to facilitating the solid–liquid separation process in such slurries. In this study, systematic experiments were conducted using a 20 kHz ultrasonic concentrator. The effects of ultrasonic treatment timing (applied at 0, 5, 10, 15, 20, 25, 30, and 35 min during free settling) and power (50 to 400 W in eight levels) were investigated by monitoring the solid–liquid interface settling velocity and underflow concentration. The key findings are as follows: Ultrasonic application at the 5 min mark yielded the optimal thickening performance, increasing the final mass concentration by 1.3% compared to free settling alone. The average settling velocity generally increased with ultrasonic power (with the exception of 50 W), and the final underflow concentration exhibited a steady rise. Notably, the 400 W treatment induced a significant settlement acceleration, attributed to the formation of drainage channels. Mechanistic analysis revealed that these drainage channels undergo a dynamic process of formation, expansion, contraction, and closure, driven by ultrasonically induced directional water migration, particle compaction, and energy boundary effects. This research not only enriches the theoretical framework of ultrasonic-assisted thickening but also provides practical insights for optimizing mine backfill operations.