XLXiaolong Li

Xiaolong Li

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Xiaolong Li has a total of 7 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • From pixels to mass gangue ratio: a lightweight DEM‐assisted end‐to‐end framework for gangue ratio in top coal caving

    by Jiahao Ma, Bo Li, Xiaolong Li, Tao Shi, Xianzhi Zou, Chenghao Zhang

    When conducting top coal caving, the timely closure of the coal discharge port is crucial for controlling the recovery rate of top coal. The existing semantic segmentation models have issues such as low degree of lightweighting, and can only provide control decisions based on the pixel gangue ratio in the discharge port. This study proposes a real-time prediction framework for end-to-end inference of the mass gangue ratio from coal flow images. Its core is a lightweight model named LEG-U-Net, which integrates knowledge distillation technology and modules such as depth separable convolution, sub-pixel convolution, global shared coordinate attention, and edge guided attention. Experiments show that compared with the teacher model, LEG-U-Net reduces the number of parameters by 87%, increases the FPS by 55%, and achieves an mIoU of 95.27%, achieving a good balance between efficiency and accuracy. Furthermore, through discrete element simulation and mathematical modeling, a quantitative mapping relationship between pixel gangue ratio, volume gangue ratio, and mass gangue ratio was established, enabling end-to-end inference from images to quality gangue ratio. This method provides a new paradigm for real-time intelligent prediction of coal flow gangue ratio, and can effectively support process control and quality optimization decisions in the coal processing process.

  • DTFRC-net: Application of a coal gangue identification model incorporating acoustic vibration information in a dual-end linked system

    by Jiahao Ma, Bo Li, Xiaolong Li, Xuewen Wang, Rui Xia

    During the process of top coal caving, by monitoring the coal discharge status of the hydraulic support through sensors, it is possible to effectively achieve a balance between insufficient coal mining and excessive coal mining. Regarding the problem that the single-source sensor monitoring technology is susceptible to environmental influence and the feature extraction is insufficient, this study proposes a dual-modal fusion model based on acoustic-vibration signals.In the vibration branch, the time-domain graphs and time-frequency graphs of the three-axis vibration signals are used as inputs, while in the acoustic signal branch, the Mel spectrum and its first and second-order differences are used as inputs. In the calculation of the Mel spectrum, an adaptive band energy weighting method is proposed to enhance the feature details of the coal gangue signal in the Mel spectrum. In the feature extraction part, a channel-spatial fusion attention is designed, and the attention of the two dimensions is dynamically fused through adaptive learning. In the feature fusion part, a bidirectional cross-adaptive gated fusion attention is designed to promote the feature interaction between the two modalities and achieve information complementation and feature fusion. At the decision level, a residual branch and an auxiliary branch are added. The residual branch is used to compensate the confidence of the output of the main branch, and the auxiliary branch participates in decision fusion to further improve the performance of the model. Experimental verification shows that the model achieves an accuracy of 98.13% on the test set. In addition, this study designs a client-server dual-end linked upper computer monitoring system for this model, achieving endto-end real-time monitoring. This study further promotes the development of dual-modal fusion technology in the industrial detection field and provides new insights and research ideas for pattern recognition based on acoustic-vibration signals.