Red mud is a substantial solid waste produced by the alumina industry, with an annual global output of more than 100 million tons, while its comprehensive utilization rate is still less than 15%. The environmental pollution caused by its extensive stockpiling has become a major issue restricting the sustainable development of the industry. This paper summarizes recent research findings on red mud recycling both domestically and internationally, and based on the development status of recycling technology, proposes three collaborative approaches for red mud treatment: source reduction through production process optimization, resource recovery of valuable components, and large-scale utilization for extensive consumption. The analysis suggests that future improvements in recovery efficiency and expansion of recovery scale for red mud will require innovations in production processes to reduce output, enhancing resource utilization efficiency through multi-element collaborative recovery, and relying on large-scale utilization to improve harmless treatment levels. These measures aim to transform red mud from “red solid waste” into a “green resource,” thereby supporting the high-quality development of the aluminum industry.
Aiming at the mechanical vibration problem caused by the coupling between the flexible deformation of the long-cantilever boom structure and the dynamic characteristics of the gear-rack transmission system during the operation of the hot metal deslagging robot, this paper focuses on conducting a systematic rigid-flexible coupled dynamic modeling and vibration characteristic analysis. By integrating multi-body dynamics and finite element analysis for co-simulation, the dynamic response of the flexible boom during the swinging phase is accurately captured. Furthermore, based on modal coordinate analysis, the vibration generation mechanism and propagation characteristics are revealed.
A set of polyacrylonitrile fibers (PANFs) with distinct anionic groups were prepared and used as supports to anchor Ru(bpy)3Cl2 through a straightforward ion-exchange procedure, affording four heterogeneous photocatalysts. The successful preparation of the anionic group-functionalized fibers and their Ru-loaded counterparts, along with the retained structural integrity, were verified by scanning electron microscopy (SEM), Fourier transform infrared spectroscopy (FTIR), X-ray diffraction (XRD), elemental analysis (EA), and X-ray photoelectron spectroscopy (XPS). The C-3 thiocyanation of indole with ammonium thiocyanate was employed as a model reaction to investigate the influence of the anionic anchoring sites on the photocatalytic performance. Notably, the phenolate-functionalized fiber PAN_P-RuF exhibited the best catalytic activity, efficiently promoting the C–S coupling of 19 indole derivatives with ammonium thiocyanate to afford 51.9–99.8% yields and the C–Se coupling of 23 indole derivatives with diphenyl diselenide to afford 44.6–99.6% yields under 405 nm irradiation at room temperature, while retaining good reusability with the yield decreasing only from 98.9% to 82.9% after five runs. Moreover, a plausible heterogeneous photocatalytic mechanism was proposed based on experimental evidence.
Low-rank, high-sulfur coal resources are abundant and constitute an essential foundation for ensuring national energy security. The compound dry separation bed offers an effective approach for the removal of inorganic mineral impurities such as high-sulfur and high-ash components, thereby promoting the clean utilization of coal. This study focuses on addressing the challenge of real-time and precise identification of coal and gangue particles within the compound dry separation bed, where densely packed particles and complex textures hinder accurate recognition. To overcome this limitation, an improved texture-aware model, Texture-Aware YOLOv13 (TAYOLOv13), is proposed. The TA-YOLOv13 architecture integrates advanced modules including This model enhances feature extraction, multi-scale fusion, and model adaptability by incorporating the second-generation version of the ConvNext series of convolutional neural network architectures - Convolution Next Version 2 (ConvNextV2), Efficient Feature Representation Network (EfficentRep), and Omni-Dimensional Dynamic Convolution (ODConv), among others. ConvNextV2, EfficientRep, and ODConv, which collectively enhance feature extraction capability, multi-scale feature fusion, and model adaptability. Experimental results demonstrate that TA-YOLOv13 achieves a precision of 97.89%, recall of 97.91%, and mAP0.5 of 98.94% on the test dataset. The overall performance indicator mAP0.5:0.95 exhibits a significant improvement of 15.5% compared to the baseline model. Heatmap visualization further verifies the model's ability to effectively capture subtle textural differences between coal and gangue under complex sorting conditions. Furthermore, an intelligent closed-loop feedback control system was established based on the proposed visual perception model. Through a fuzzy PID controller, the system dynamically regulates key operational parameters such as airflow rate, vibration amplitude, and frequency. Under the optimal parameter combination (frequency 24-26 Hz, amplitude 2.2-2.6 mm, and air velocity 1.80-2.20 m/s), the intelligent dry separation of low-rank high-sulfur coal achieved remarkable separation efficiency: the clean coal contamination rate was below 2%, the clean coal loss rate was below 4%, and the clean coal yield reached 70.02% with ash content of 9.36% and sulfur content of 0.98%. The gangue yield was 29.98%, with ash content of 49.08% and sulfur content of 9.92%. Distinct gradient distributions of sulfur and ash content were observed across the bed, and the separation precision, represented by the Ep = 0.055 g/cm3. Overall, the proposed method realizes clean, efficient, and high-value upgrading of low-rank high-sulfur coal through the synergistic integration of accurate perception, intelligent decision-making, and optimized process control.
This study presents a novel methodology that combines machine learning with first-principles calculations to efficiently screen austenitic stabilizing elements. Correlation analysis identified the lattice constant as a critical factor influencing the stability of austenitic doping systems. Subsequently, the random forest, support vector regression, and AdaBoost models were evaluated, among which the random forest achieved the highest prediction accuracy ( R² = 0.748). Furthermore, SHapley Additive exPlanations was employed to interpret the model, further verifying the potential role of the lattice constant. Based on the above studies, Cr, Ni, and Mn were identified as doping elements. Finally, the first-principles calculation is employed to verify the prediction results of the machine learning. First-principles calculation results revealed that the austenite doped with Cr exhibits the lowest system energy (-29609.59 eV) and solid solution energy (-11.795 eV). Electronic structure analysis (including charge density difference and density of states) reveals the underlying mechanism: the larger the lattice constant of the doping atom, the weaker its interaction with the iron atoms, as specifically manifested by the reduction in electron cloud density and covalent bonding. These findings not only confirm the scientific validity of the proposed integrated machine learning and first-principles approach but also offer important guidance for the rational design of high-performance austenitic materials.
