深部金属矿山 TBM 智能掘进感知与决策框架研究
朱梦琦 , 李建斌 , 朱合华 , 刘永胜 , 姚诞圣 , 白孟琦 , 孙诗尧
有色金属(矿山部分) ›› 2026, Vol. 78 ›› Issue (6) : 111 -123.
深部金属矿山 TBM 智能掘进感知与决策框架研究
A perception and decision-making framework for intelligent TBM excavation in deep metal mines
随着深部金属矿山开采向高地应力、高地温及复杂地质环境延伸,隧道掘进机(TBM)施工过程面临多源异构数据融合困难、复杂工况感知不足,以及施工决策高度依赖人工经验等挑战。与交通隧道工程不同,深部金属矿山巷道掘进遭遇的围岩条件变化更为剧烈且频繁,矿岩界面交替、高应力诱导的岩爆倾向以及多灾害耦合效应,使TBM施工过程具有极强的非线性和动态不确定性。现有研究多聚焦于参数预测、地质识别或风险预警等单一任务,尚未形成贯穿数据底层融合、状态信息提取、演化规律认知与施工决策生成的统一智能体系。针对上述问题,本文从智能体视角出发,提出一种面向深部金属矿山TBM 智能掘进的感知与决策框架。该框架构建了逐层递进的四个核心层次:数据认知层基于“人- 机- 岩”耦合机理,实现多源数据融合与统一特征表达;状态感知层融合知识驱动与数据驱动方法,实现装备状态、地质状态及风险状态的快速识别;时空分析层揭示围岩条件和设备响应的时空演化规律,实现边界识别、可掘性表征及未来运行响应预测;决策生成层依托大语言模型、知识图谱及多维证据融合机制,生成风险预警、参数优化及施工决策。最终形成“数据认知- 状态感知- 时空分析- 决策生成”的闭环,实现了由数据驱动向知识- 数据协同驱动、由被动响应向主动前瞻决策的范式转变,为深部金属矿山TBM 智能掘进提供可行的理论框架和技术路径。
As deep metal mining advances toward environments characterized by high in-situ stress,elevated ground temperature,and complex geological conditions,Tunnel boring machine (TBM) excavation faces critical challenges,including difficulties in multi-source heterogeneous data fusion,insufficient perception of complex operating conditions,and excessive dependence on human experience for construction decision-making. Unlike conventional transportation tunnel projects,deep metal mine roadways experience more drastic and frequent variations in surrounding rock conditions. The alternating occurrence of ore-rock interfaces,high-stress-induced rockburst potential,and coupled multi-hazard effects introduce strong nonlinearity and dynamic uncertainty into TBM excavation processes. Existing studies mainly focus on individual tasks,such as performance prediction,geological identification,or risk warning,and have not yet established a unified intelligent system that integrates underlying data fusion,state information extraction,evolutionary mechanism understanding,and construction decision generation. To address these challenges,this paper proposes an agent-oriented perception and decision-making framework for intelligent TBM excavation in deep metal mines. The proposed framework consists of four progressively hierarchical layers:the Data Cognition Layer,which integrates multi-source data and achieves unified feature representation based on the human-machine-rock coupling mechanism;the State Perception Layer,which combines knowledge-driven and data-driven approaches to enable rapid identification of equipment states,geological states,and risk states;the Spatiotemporal Analysis Layer,which uncovers the spatiotemporal evolution patterns of surrounding rock conditions and machine responses,enabling boundary detection,boreability characterization,and future operational response prediction;and the Decision Generation Layer,which leverages large language models,knowledge graphs,and multi-dimensional evidence fusion mechanisms to generate risk warnings,parameter optimization strategies,and intelligent construction decisions. Ultimately,the proposed framework establishes a closed-loop architecture spanning data cognition,state perception,spatiotemporal analysis,and decision generation,enabling a paradigm shift from data-driven intelligence toward knowledge-data collaborative intelligence,and from passive response toward proactive and forward-looking decision-making. This framework provides a feasible theoretical foundation and technical pathway for the development of intelligent TBM excavation systems in deep metal mines.
| [1] |
蔡美峰,薛鼎龙,任奋华. 金属矿深部开采现状与发展战略[J]. 工程科学学报,2019,41(4):417-426. |
| [2] |
|
| [3] |
刘泉声,潘玉丛,余宏淦, |
| [4] |
|
| [5] |
黄丹,杨小聪,陈何. TBM 在地下金属矿山应用中的发展现状与趋势[J]. 有色金属工程,2019,9(11):108-121. |
| [6] |
|
| [7] |
|
| [8] |
|
| [9] |
|
| [10] |
|
| [11] |
周小雄,龚秋明,殷丽君, |
| [12] |
|
| [13] |
|
| [14] |
|
| [15] |
|
| [16] |
|
| [17] |
|
| [18] |
|
| [19] |
|
| [20] |
|
| [21] |
|
| [22] |
|
| [23] |
|
| [24] |
李建斌,武颖莹,李鹏宇, |
| [25] |
|
| [26] |
|
| [27] |
|
| [28] |
Rostami J. Development of a force estimation model for rock fragmentation with disc cutters through theoretical modeling and physical measurement of crushed zone pressure[D]. Golden:Colorado School of Mines,1997:56. |
| [29] |
|
| [30] |
|
| [31] |
|
| [32] |
|
| [33] |
|
国家自然科学基金资助项目(52408437)
/
| 〈 |
|
〉 |