College of Computer Science and Technology, Xi’an University of Science and Technology
Hebei Key Laboratory of Mine Intelligent Unmanned Mining Technology, North China Instituteof Science and Technology
College of Engineering, Xi’ an International University

工作面区域液压支架压力变化会导致支架关联区域安全态势发生变化,提前预测支架压力值变化能够在一定程度上预防相应的事故发生。聚焦煤矿工作面支架下的人员安全问题,根据支架压力和人员轨迹的预测结果,提前对移动轨迹可能经过危险支架位置的人员进行提醒,提高对工作面区域作业人员的安全保障水平。首先,提出了基于AMG-TSMixer支架群组的压力值预测方法,通过AMG算法充分挖掘同一工作面不同支架的时空关系,利用TSMixer来预测支架群组的压力值,结合矿方现有的矿压监测预警系统和方法,对支架所处区域的安全状态进行超前分析;其次,提出了一种基于GRU-Transformer网络的人员轨迹预测方法,通过GRU网络对时空特征进行捕捉,从而增强Transformer模型的记忆能力,把人员之间的复杂相互作用关系进行更详细的挖掘。通过人员轨迹和支架安全区域的分析判断结果,给人员安全保障提供未来的解决方案。结果表明:所提出的AMG-TSMixer方法在RMSE和MAPE两项指标上相比于同类代表性方法提升了5.9%,10.7%;GRU-Transformer方法在ADE和FDE两项指标上相比于同类代表性方法提升了14%和19%。支架群组的压力值预测和人员轨迹预测方法相互结合,不仅能有效预测出支架群组的压力值,还能预测出更真实的人员运动轨迹,对煤矿综采工作面上作业人员的安全保障发挥了重要作用。

The change of the hydraulic support pressure in the working face area can lead to the change of the safety situation in the support associated area. Predicting the change of support pressure values in advance can prevent the corresponding accidents to a certain extent. The paper focused on the research of personnel safety guarantee method under the support of coal mine working face. According to the prediction results of support pressure and personnel trajectory, the personnel whose moving trajectory may pass through the dangerous support position were reminded in advance, so as to improve the safety guarantee level of workers in the working face area. Firstly, a mine pressure value of the support groups prediction algorithm based on AMG-TSMixer was proposed. The AMG algorithm was used to fully exca- vate the spatial-temporal relationship of different supports in the same working face, and TSMixer was used to predict the mine pressure value of the support groups. Combined with the existing mine pres- sure monitoring and early warning system and method, the safety state of the support area was analyzed in advance. Secondly, a personnel trajectory prediction algorithm based on the GRU-Transformer net- work was proposed, which captures spatial-temporal features through the GRU network, thereby enhan- cing the memory ability of the Transformer model and excavating the complex interaction relationships between personnel in more detail. Through the analysis and judgment results of personnel trajectory and support safety area, it provided future solutions for personnel safety guarantee. The results show that the proposed AMG-TSMixer method has improved by 5. 9% and 10. 7% compared with similar representa- tive methods in the two indicators of RMSE and MAPE, and can effectively predict the mine pressure value of the support groups; The proposed GRU-Transformer method in this paper has improved by 14% and 19% compared with similar representative methods in the two indicators of ADE and FDE, and can predict more realistic personnel trajectory. The results show that the proposed AMG-TSMixer method has improved by 5. 9% and 10. 7% compared with similar representative methods in terms of RMSE and MAPE. The proposed GRU-Transformer method in this paper has improved by 14% and 19% compared with similar representative methods in the two indicators of ADE and FDE. The combi- nation of the pressure value prediction of the support group and the personnel trajectory prediction method can not only effectively predict the pressure value of the support group, but also predict a more realistic personnel trajectory, which plays an important role in the safety guarantee of the workers on the fully mechanized mining face of the coal mine.

国家自然科学基金项目(52274137);河北省矿山智能化开采技术重点实验室开放基金项目(iium004);陕西省教育厅专项科研计划项目(22JK0459);陕西省“十四五”教育科学规划项目(SGH22Y1883)第一作者:张婧,女,陕西西安人,博士,副教授,E-mail:zhangjing21@xust.edu.cn

张婧,李冰,张科学,等.综采工作面液压支架压力预测下人员安全保障研究[J].西安科技大学学报,2024,44(2):379-390.

ZHANG Jing, LI Bing, ZHANG Kexue, et al. Personnel safety guarantee under pressure prediction of hydraulic support in fully mech-anized mining face[ J] . Journal of Xi’an University of Science and Technology, 2024, 44( 2) : 379-390.