矿井瓦斯抽采管网的传统调控方式主要依赖人工经验和固定规则，难以适应复杂多变的井下环境，存在瓦斯抽采效率低、能耗高，甚至发生安全隐患的问题。近年来随着煤矿智能化建设工作的推进，矿井瓦斯抽采管网系统的信息化与智能化也成为矿井发展的必然趋势，本课题以陕西亭南煤矿为研究试验点，对瓦斯抽采管网的智能化调控开展了以下研究工作：

（1）理论分析与调控框架设计：将瓦斯抽采管网可抽象为有向图结构，基于基尔霍夫定律与哈迪-克罗斯法建立质量-能量守恒方程，为后续流量调度与调控策略提供了结构化的数学表达基础；量化流量分配与压力损失规律用来描述管网中的流量分配与压力变化规律；对设备调节分析，得到了阀门开度大小和抽采泵功率对调节管网负压，瓦斯纯流量的影响。安全约束上，需严格满足CO浓度、抽采负压、抽采温度等硬性阈值；效率约束上，要求瓦斯浓度与流量持续高于临界值，为深部矿井安全开采提供了理论支撑与技术规范。同时介绍了陕西亭南煤矿的瓦斯抽采管网系统概括，包括主要监测节点的布置和系统组成，结合相关国家煤矿瓦斯抽采智能化建设标准和具体规范，构建实现瓦斯抽采管网智能调控方案的理论框架。

（2）瓦斯浓度预测模型的构建和验证：分析了瓦斯抽采管网的时空特性，构建了基于双重注意力机制的时空图神经网络（DASTNN）预测模型。该模型将瓦斯管网拓扑结构与瓦斯浓度历史数据作为输入，结合图卷积网络（GCN）提取空间特征、门控循环单元（GRU）提取时间序列特征，并引入时空注意力机制进行加权学习，实现对管网各节点瓦斯浓度的精准预测。实验结果表明：在瓦斯浓度预测任务中，DASTNN模型相较于传统GRU模型，MAE降低0.239，RMSE降低0.821，R²提升0.043，验证了其预测性能的优越性与稳定性。

（3）瓦斯抽采管网优化调控策略的建模和求解：在明确瓦斯浓度、瓦斯纯流量和抽采泵效率作为调控目标的基础上，针对抽采负压与瓦斯浓度、纯流量之间的不匹配和抽采效率低的问题，构建了两级优化调度模型：具体为一级模型以获取最大瓦斯浓度和纯流量为目标，二级模型进一步优化阀门开度与抽采泵功率。为求解该模型，提出结合粒子群优化（PSO）与近端策略优化（PPO）的PSO-PPO智能调控算法，实现对阀门与抽采泵功率的动态最优调节，使系统运行在高效、安全的最优状态。实验结果表明：在调控优化方面，PSO-PPO算法在抽采泵站功率上实现了最大69.7kW的降幅，抽采效率最大提升了13.3%。

The traditional control method of mine gas extraction pipeline network mainly relies on manual experience and fixed rules, which is difficult to adapt to the complex and changeable underground environment. There are problems such as low gas extraction efficiency, high energy consumption, and even safety hazards. In recent years, with the advancement of coal mine intelligent construction, the informatization and intelligence of mine gas extraction pipeline network system has also become an inevitable trend in the development of mines. This project takes a coal mine in Changwu, Shaanxi as a research and experimental point, and carries out the following research work on the intelligent control of gas extraction pipeline network:

(1) Theoretical analysis and control framework design: The gas extraction network can be abstracted as a directed graph structure, and the mass-energy conservation equation is established based on Kirchhoff's law and Hardy-Cross method, which provides a structured mathematical expression basis for subsequent flow scheduling and control strategies; the quantitative flow distribution and pressure loss law are used to describe the flow distribution and pressure change law in the network; the equipment adjustment analysis obtains the influence of valve opening size and extraction pump power on regulating the network negative pressure and pure gas flow. In terms of safety constraints, hard thresholds such as CO concentration, extraction negative pressure, and extraction temperature must be strictly met; in terms of efficiency constraints, the gas concentration and flow rate are required to be continuously higher than the critical value, which provides theoretical support and technical specifications for safe mining in deep mines. At the same time, the gas extraction network system of Tingnan Coal Mine in Shaanxi is introduced, including the layout of the main monitoring nodes and the system composition. Combined with the relevant national coal mine gas extraction intelligent construction standards and specific specifications, a theoretical framework for realizing the intelligent control scheme of the gas extraction network is constructed.

(2) Construction and verification of gas concentration prediction model: The spatiotemporal characteristics of the gas extraction network are analyzed, and a spatiotemporal graph neural network (DASTNN) prediction model based on the dual attention mechanism is constructed. The model takes the gas network topology and gas concentration historical data as input, combines the graph convolutional network (GCN) to extract spatial features, the gated recurrent unit (GRU) to extract time series features, and introduces the spatiotemporal attention mechanism for weighted learning to achieve accurate prediction of gas concentration at each node of the network. The experimental results show that in the gas concentration prediction task, compared with the traditional GRU model, the DASTNN model has a MAE reduction of 0.239, a RMSE reduction of 0.821, and an R² increase of 0.043, which verifies the superiority and stability of its prediction performance.

(3) Modeling and solving the optimization control strategy of the gas extraction pipeline network: On the basis of clarifying the gas concentration, pure gas flow rate and extraction pump efficiency as the control targets, a two-level optimization scheduling model is constructed to address the mismatch between the extraction negative pressure and the gas concentration and pure flow rate and the low extraction efficiency: the first-level model aims to obtain the maximum gas concentration and pure flow rate, and the second-level model further optimizes the valve opening and the extraction pump power. In order to solve the model, a PSO-PPO intelligent control algorithm combining particle swarm optimization (PSO) and proximal strategy optimization (PPO) is proposed to achieve dynamic optimal adjustment of valves and extraction pump power, so that the system can operate in an efficient and safe optimal state. The experimental results show that in terms of regulation and optimization, the PSO-PPO algorithm achieved a maximum reduction of 69.7kW in the power of the extraction pump station, and the extraction efficiency was increased by up to 13.3%.

[1] 刘程, 孙东玲, 武文宾, 等. 我国煤矿瓦斯灾害超前大区域精准防控技术体系及展望[J]. 煤田地质与勘探, 2022, 50(8): 82-92.

[2] Liu J, Wang Z, Wang Q, et al. Study on the application of comprehensive gas control technology for Source-Separated Three-Dimensional gas extraction in large mining faces[J]. ACS omega, 2024, 9(30): 32777-32788.

[3] Zhao C, Gao P, Ruan J, et al. Research on Hotspots and Evolutionary Trends in Coal Mine Gas Prevention[J]. Processes, 2024, 12(9): 1993.

[4] Yang F, Ge Z, Chen J, et al. A comprehensive gas extraction system coupling high-level suction roadway and boreholes for gas disaster prevention in closely-spaced multiple coal seams[J]. Energy Sources, Part A: Recovery, Utilization, and Environmental Effects, 2025, 47(1): 2302-2315.

[5] Yuxin W, Gui F, Qian L, et al. Accident case-driven study on the causal modeling and prevention strategies of coal-mine gas-explosion accidents: A systematic analysis of coal-mine accidents in China[J]. Resources Policy, 2024, 88: 104425.

[6] Jiang C, Zeng Y, Deng B, et al. Mechanisms underlying variations in pore structures and permeability in response to thermal treatment and their implications for underground coal utilization[J]. Powder Technology, 2024, 444: 120073.

[7] 倪坤, 刘闯, 王妍. “十四五”时期世界煤炭工业发展现状及趋势研究[J]. 中国煤炭, 2024, 50(8):235-245.

[8] 赵晓斌. 煤矿瓦斯爆炸事故的防治对策研究[J]. 中国高新技术企业, 2015, (20): 150-151.

[9] 戴春雷.我国矿山智能化研究进展及大模型应用前景[J].工矿自动化,2024,50(S2):111-133.

[10] 徐刚,牛航宇,赵海波,等.复杂地质条件下近距离煤层群开采瓦斯涌出特征及治理技术研究[J].矿业研究与开发,2024,44(08):157-164.

[11] 李成武,乔朕.煤与瓦斯突出局部防突措施失效判定方法[J].煤炭学报,2024,49(S2):924-933.

[12] 程成,成小雨,周爱桃,等.智能瓦斯抽采负压的动态调控策略及效果研究[J].中国矿业,2025-6-13.

[13] 周福宝,刘春,夏同强,等.煤矿瓦斯智能抽采理论与调控策略[J].煤炭学报,2019,44(08):2377-2387.

[14] Lu Z, Wang L, Fu S, et al. High-efficiency and precision gas extraction in intelligent mining faces: Application of a comprehensive three-dimensional extraction system[J]. Physics of Fluids, 2024, 36(8).

[15] Qin Pei, Lian Changbao, Li Xiyuan, et al. Modeling technology and intelligent implementation of refined design for cross-layer boreholes[J]. Coal Mine Safety, 2025, 56(01): 220–227.

[16] Wan Y, Xin D, Wan J, et al. Prospect of Cluster Control Technology for Automatic Gas Extraction Drilling Rig in Coal Mine[J]. Academic Journal of Engineering and Technology Science, 2024, 7(4): 72-78.

[17] SONG Xi, SHU Longyong, WANG Bi, et al. Replacement and pumpig technology for coal seams with low gas content and high mtensity miningly Joumal of lining and Safety Engineering. 2023.40(4):847-856.

[18] 李国富,李超,张碧川,等.我国煤矿瓦斯抽采与利用发展历程、技术进展及展望[J].煤田地质与勘探,2025,53(01):77-91.

[19] 张慧杰,霍中刚,舒龙勇,等.煤矿瓦斯抽采系统智能管控研究现状及展望[J].煤炭科学技术,2025-6-13.

[20] 李小军,赵明炀,李淼.基于深度学习的钻孔冲煤量智能识别方法[J].煤田地质与勘探,2025,53(01):257-270.

[21] 郭世斌,胡国忠,朱家锌,等.顶板瓦斯抽采巷布置位置智能预测方法[J].煤炭科学技术,2024,52(04):203-213.

[22] 郝天轩,范国阳,赵立桢.基于AutoCAD的钻孔智能设计系统[J].中国安全生产科学技术,2023,19(04):71-77.

[23] 郭恒,王昊,程晓阳.基于数据驱动的瓦斯抽采钻孔精细化管控[J].矿业安全与环保,2022,49(03):125-130.

[24] Yang G, Zhu Q, Wang D, et al. Method and validation of coal mine gas concentration prediction by integrating PSO algorithm and LSTM network[J]. Processes, 2024, 12(5): 898.

[25] HU Q, ZHENG S, LI S, et al. An intelligent prediction method of gas concentration in coal mines based onimproved TCN-TimeGAN[J]. COAL SCIENCE AND TECHNOLOGY, 2024, 52(S2): 321-330.

[26] Jia P, Liu H, Wang S, et al. Research on a mine gas concentration forecasting model based on a GRU network[J]. IEEE access, 2020, 8: 38023-38031.

[27] Zhang S, Wang B, Li X, et al. Research and application of improved gas concentration prediction model based on grey theory and BP neural network in digital mine[J]. Procedia Cirp, 2016, 56: 471-475.

[28] Yang G, Zhu Q, Wang D, et al. Method and validation of coal mine gas concentration prediction by integrating PSO algorithm and LSTM network[J]. Processes, 2024, 12(5): 898.

[29] Tutak M, Krenicky T, Pirník R, et al. Predicting methane concentrations in underground coal mining using a multi-layer perceptron neural network based on mine gas monitoring data[J]. Sustainability, 2024,16(19): 8388.

[30] 梁运培,李赏,李全贵,等.基于FEDformer-LGBM-AT架构的采煤工作面上隅角瓦斯浓度预测[J].煤炭学报,2025,15(5):1-18.

[31] 李玉鹏,贾澎涛,刘平,等.基于MHA-BiGRU的工作面瓦斯浓度预测模型[J].中国矿业,2024,33(12):180-188.

[32] 曹亮.基于Adam算法BP神经网络的矿井环境瓦斯浓度预测模型[J].湖南科技大学学报(自然科学版),2024,39(01):18-23.

[33] 龚晓燕,孙育恒,刘壮壮,等.基于BP神经网络的煤矿综掘面风流智能调控模型研究[J].安全与环境工程,2023,30(06):138-145.

[34] 兰永青,乔元栋,程虹铭,等.基于SSA-LSTM的瓦斯浓度预测模型[J].工矿自动化,2024,50(02):90-97.

[35] 胡青松,郑硕,李世银,等.基于改进TCN-TimeGAN的矿井瓦斯浓度智能预测方法[J].煤炭科学技术,2024,52(S2):321-330.

[36] 秦岩,盛武.基于PCA-GRU-SVM模型的多参量煤矿瓦斯浓度预测及预警研究[J].安全与环境工程,2023,30(06):81-88.

[37] 杨科,宋克静,张杰.基于多尺度卷积和注意力机制的MCA-GRU瓦斯浓度预测模型[J].矿业研究与开发,2023,43(10):154-159.

[38] 曹梅,杨超宇.基于小波的CNN-LSTM-Attention瓦斯预测模型研究[J].中国安全生产科学技术,2023,19(09):69-75.

[39] Zhou F, Xia T, Wang X, et al. Recent developments in coal mine methane extraction and utilization in China: a review[J]. Journal of Natural Gas Science and Engineering, 2016, 31: 437-458.

[40] Zhou F, Liu C, Xia T, et al. Intelligent gas extraction and control strategy in coal mine[J]. Journal of China Coal Society, 2019 (8): 2377-2387.

[41] 齐黎明,王声远,陈学习,等.我国煤矿瓦斯抽采智能化技术现状与展望[J].华北科技学院学报,2023,20(05):71-75.

[42] 李泉新,程卓尔,方俊,等.定向长钻孔瓦斯抽采负压变化规律及监测控制技术研究进展[J].煤田地质与勘探,2024,52(11):171-182.

[43] 李鹏,王衍海,周爱桃,等.煤层瓦斯抽采主控因素演化机制研究[J].煤炭工程,2024,56(08):165-171.

[44] 高亚斌,刘国强,张伟浩,等.高瓦斯涌出工作面反向孔联通高抽巷瓦斯治理技术[J].煤炭科学技术, 2025-6-13.

[45] 潘吉成.基于三维流固耦合模型的煤层瓦斯有效抽采半径研究[J].中国煤炭,2024,50(S1):128-134.

[46] 蔡峰.王家岭煤矿综放工作面上覆岩层运动规律及卸压区瓦斯抽采试验研究[J].中国煤炭,2024,50(01):42-51.

[47] 金元甲,马凯,张亚洲,等.定向孔抽采特性对布孔参数响应变化规律研究[J].矿业安全与环保,2023,50(06):120-124.

[48] 龚选平,周爱桃,范席辉,等.基于“管网解算-遗传算法寻优”的瓦斯抽采管网优化技术[J].煤矿安全,2022,53(10):120-125.

[49] 李生亚,高坤,张斌峰,等.瓦斯抽采泵站智能化应用研究[J].中国安全科学学报,2023,33(S2):85-90.

[50] 马莉,石新莉,李树刚,等.基于MPC的瓦斯抽采智能调控模型研究[J].煤炭科学技术,2022,50(08):82-90.

[51] 张增辉,邓飞,黄友胜.瓦斯抽采智能管控平台研究与应用[J].中国煤炭,2025,51(03):93-100.

[52] 昝小舒,池宇洋,夏同强,等.矿井瓦斯抽采参数调控实验教学系统研制[J].实验室研究与探索,2024,43(06):29-33.

[53] Lu B, Gan X, Jin H, et al. Make more connections: Urban traffic flow forecasting with spatiotemporal adaptive gated graph convolution network[J]. ACM Transactions on Intelligent Systems and Technology (TIST), 2022, 13(2): 1-25.

[54] 谢娟英,张建宇.图卷积神经网络综述[J].陕西师范大学学报(自然科学版),2024,52(02):89-101.

[55] 吴博,梁循,张树森,等.图神经网络前沿进展与应用[J].计算机学报,2022,45(01):35-68.

[56] Jiang W, Luo J. Graph neural network for traffic forecasting: A survey[J]. Expert Systems with Applications, 2022, 207: 117921.

[57] 温志勇,翁小雄,谢帮权.基于多层复杂网络的循环神经网络交通量预测模型[J].现代电子技术,2024,47(22):173-178.

[58] 邓斌,张楠,王江,等.基于LTC-RNN模型的中长期电力负荷预测方法[J].天津大学学报(自然科学与工程技术版),2022,55(10):1026-1033.

[59] 刘祥祥,朱一松,高越飞,等.改进粒子群算法耦合水力模型的供水优化调度分析[J].中国给水排水,2025,41(09):59-65.

[60] 张丽,刘青雷,艾恒涛,等.基于需求响应的家庭负荷优化调度算法研究[J].太阳能学报,2024,45(09):131-139.

[61] 梁涛,张晓婵,谭建鑫,等.基于PPO算法的电热氢耦合综合能源系统优化调度[J].太阳能学报,2024,45(11):73-83.

[62] 张黎元,宋兴旺,李冰洁,等.基于自注意力PPO算法的智能配电网多设备协同无功优化控制策略[J].智慧电力,2024,52(10):40-48.

[63] 付华, 付昱, 赵俊程, 等.基于KPCA-ARIMA算法的瓦斯涌出量预测[J]. 辽宁工程技术大学学报(自然科学版), 2022, 41(5):406-412.

[64] Dong D. Mine gas emission prediction based on Gaussian process model[J]. Procedia Engineering, 2012, 45: 334-338.

[65] Bi S, Shao L, Qi Z, et al. Prediction of coal mine gas emission based on hybrid machine learning model[J]. Earth Science Informatics, 2023, 16(1): 501-513.

[66] Liang R, Chang X, Jia P, et al. Mine gas concentration forecasting model based on an optimized BiGRU network[J]. ACS omega, 2020, 5(44): 28579-28586.

[67] LI S, MA L, PAN S, et al. Research on prediction model of gas concentration based on RNN in coal mining face[J]. Coal Science and Technology, 2020, 48(1).

[68] Noack K. Control of gas emissions in underground coal mines[J]. International Journal of Coal Geology, 1998, 35(1-4): 57-82.

[69] 张超林,刘明亮,王恩元,等.煤层渗透性对煤与瓦斯突出的影响规律及控制机理[J].煤炭学报,2024,49(12):4842-4854.

[70] Moran Z H U. 3D simulation and control effect evaluation of gas extraction drilling in coal mine[J]. Mining Safety & Environmental Protection, 2024, 51(3): 50-55.

[71] 熊伟.基于图论的瓦斯抽采管网优化技术[J].煤矿安全,2020,51(04):88-92.

[72] Wei P, Huang C, Li X, et al. Numerical simulation of boreholes for gas extraction and effective range of gas extraction in soft coal seams[J]. Energy Science & Engineering, 2019, 7(5): 1632-1648.

[73] He L, Dai Y, Xue S, et al. Study on gas control methods optimization for mining safety[J]. Advances in Civil Engineering, 2021, 2021(1): 4594330

[74] 徐宝昌,杨青松,吴一凡,等.面向采出水集输系统的两级优化调度[J].给水排水,2024,60(S1):721-727.