基于MTAM-LSTM的采煤工作面支架载荷预测方法*

doi:10.16265/j.cnki.issn1003-3033.2026.03.1821

摘要/Abstract

摘要：

为有效预测液压支架载荷、评估支架运行状态,提出一种基于多尺度卷积时间注意力模块(MTAM)-长短时记忆(LSTM)神经网络的液压支架载荷预测模型。首先,采用自适应噪声完备集合经验模态分解(CEEMDAN)算法分解支架载荷数据获取本征模态分量,基于K-L散度准则剔除本征模态分量中的冗余分量形成支架载荷预测输入序列;其次,建立MTAM捕捉支架载荷变化特征,静态注意力生成数据特征信息的注意力权重,动态注意力优化不同序列特征的关注度,并引入残差学习保持特征信号的完整性;然后,利用LSTM构建特征信息与支架载荷之间的深层依赖关系,实现支架载荷数据的超前预测;最后,选取陕西某冲击地压矿井402102工作面液压支架载荷数据进行实证分析,对比不同模型均方根误差、决定系数和平均绝对误差3种评价指标,结果表明:MTAM-LSTM模型的均方根误差(RMSE)和平均绝对误差(MAE)均明显小于对比模型,RMSE整体降低0.16~0.45,MAE降低0.16~0.45,不同场景下决定系数R²达到0.91,验证了MTAM-LSTM的预测准确率和模型泛化能力。

关键词: 多尺度卷积时间注意力模块(MTAM), 长短时记忆神经网络(LSTM), 采煤工作面, 载荷预测, 液压支架, 泛化能力

Abstract:

In order to effectively predict hydraulic support loads and evaluate the operational status of supports, a hydraulic support load prediction model based on MTAM-LSTM was proposed. The CEEMDAN algorithm was employed to decompose the load data of supports and extract intrinsic mode functions. Redundant components in the intrinsic mode functions were eliminated according to K-L divergence criterion, thereby forming the input sequence for load prediction. An MTAM was constructed to capture the variation characteristics of hydraulic support loads. Static attention generated attention weights for feature information of data, while dynamic attention optimized the focus on different sequence features. Residual learning was introduced to maintain the integrity of feature signals. LSTM networks were then utilized to establish deep dependencies between feature information and hydraulic support loads, enabling advanced prediction of support load data. Field data from the 402102 working face of a rockburst-prone coal mine in Shaanxi were used for empirical validation. RMSE, R², and MAE were used as evaluation metrics for comparison among different models. The results show that the RMSE and MAE of the MTAM-LSTM model are significantly lower than those of the comparison models, with RMSE reduced by 0.16-0.45 and MAE reduced by 0.16-0.45, while the coefficient of determination R² reaches 0.91 under different scenarios, thereby validating the prediction accuracy and generalization capability of MTAM-LSTM model.

Key words: multi-convolution temporal attention module (MTAM), long short-term memory (LSTM), mining working face, load prediction, hydraulic support, generalization ability

中图分类号:

X936
TD 323

张杰, 杨科, 范超尘. 基于MTAM-LSTM的采煤工作面支架载荷预测方法^*[J]. 中国安全科学学报, 2026, 36(3): 144-152.

ZHANG Jie, YANG Ke, FAN Chaochen. Prediction method of support load in coal mining face based on MTAM-LSTM[J]. China Safety Science Journal, 2026, 36(3): 144-152.

图/表 14

图1

图2

表1

图3

表2

图4

图5

表3

图6

图7

表4

图8

表5

表6

参考文献 23

[1]	庞义辉, 王国法, 任怀伟. 大采高工作面煤壁片帮的多因素敏感性分析[J]. 采矿与安全工程学报, 2019, 36(4): 736-745.
	PANG Yihui, WANG Guofa, REN Huaiwei. Multiple influence factor sensitivity analysis on coal wall spalling of workface with large mining height[J]. Journal of Mining & Safety Engineering, 2019, 36(4): 736-745.
[2]	王慧, 赵国超, 宋宇宁, 等. 基于改进的威布尔分布的液压支架可靠性评估方法[J]. 中国安全科学学报, 2018, 28(5): 99-104. doi: 10.16265/j.cnki.issn1003-3033.2018.05.017
	WANG Hui, ZHAO Guochao, SONG Yuning, et al. Reliability evaluation method of hydraulic support based on improved Weibull distribution[J]. China Safety Science Journal, 2018, 28(5): 99-104. doi: 10.16265/j.cnki.issn1003-3033.2018.05.017
[3]	魏文艳. 综采工作面智能化开采技术发展现状及展望[J]. 煤炭科学技术, 2022, 50(增2): 244-253.
	WEI Wenyan. Development status and prospect of intelligent mining technology of longwall mining[J]. Coal Science and Technology, 2022, 50(S2): 244-253.
[4]	杨科, 刘千贺, 李志华. 厚松散层上提工作面覆岩运移与支架-围岩关系研究[J]. 煤炭科学技术, 2015, 43(10): 12-17.
	YANG Ke, LIU Qianhe, LI Zhihua. Study on relationship between overburden strata migration in thick loose strata with extending upper extraction limit coal mining face and support-surrounding rock[J]. Coal Science and Technology, 2015, 43(10): 12-17.
[5]	刘畅, 刘正和, 张俊文, 等. 工作面长度对覆岩空间结构演化及大采高采场矿压规律的影响[J]. 岩土力学, 2018, 39(2): 691-698.
	LIU Chang, LIU Zhenghe, ZHANG Junwen, et al. Effect of mining face length on the evolution of spatial structure of overlying strata and the law of underground pressure in large mining height face[J]. Rock and Soil Mechanics, 2018, 39(2): 691-698.
[6]	李云鹏, 赵善坤, 李杨, 等. 复杂坚硬岩层条件下特厚煤层综放开采矿压分级预测[J]. 煤炭学报, 2021, 46(增1): 38-48.
	LI Yunpeng, ZHAO Shankun, LI Yang, et al. Prediction on weighting classification of fully-mechanized caving mining under extremely thick coal seam[J]. Journal of China Coal Society, 2021, 46(S1): 38-48.
[7]	曾庆田, 吕珍珍, 石永奎, 等. 基于Prophet+LSTM模型的煤矿井下工作面矿压预测研究[J]. 煤炭科学技术, 2021, 49(7): 16-23.
	ZENG Qingtian, LYU Zhenzhen, SHl Yongkui, et al. Research on prediction of underground coal mining face pressure based on Prophet+LSTM model[J]. Coal Science and Technology, 2021, 49(7): 16-23.
[8]	霍丙杰, 荆雪冬, 于斌, 等. 坚硬顶板厚煤层采场来压强度分级预测方法研究[J]. 岩石力学与工程学报, 2019, 38(9): 1828-1835.
	HUO Bingjie, JING Xuedong, YU Bin, et al. A classification forecasting method for the weighting intensity of stopes of thick coal seams with hard roofs[J]. Chinese Journal of Rock Mechanics and Engineering, 2019, 38(9): 1828-1835.
[9]	徐刚, 张春会, 蔺星宇, 等. 基于分区支承力学模型的综放工作面顶板矿压演化与压架预测[J]. 煤炭学报, 2022, 47(10): 3622-3633.
	XU Gang, ZHANG Chunhui, LIN Xingyu, et al. Predicting ground pressure evolution and support crushing of fully mechanized top coal caving face based on zoning support mechanical model[J]. Journal of China Coal Society, 2022, 47(10): 3622-3633.
[10]	贺超峰, 华心祝, 杨科, 等. 基于BP神经网络的工作面周期来压预测[J]. 安徽理工大学学报:自然科学版, 2012, 32(1): 59-63.
	HE Chaofeng, HUA Xinzhu, YANG Ke, et al. Forecast of periodic weighting in working face based on back-propagation neural network[J]. Journal of Anhui University of Science and Technology:Natural Science, 2012, 32(1): 59-63.
[11]	XIAO Yigai, DENG Hongwei, XIE Zhimou, et al. Mine ground pressure monitoring and early warning based on deep learning data analysis[J]. Mobile Information Systems, 2022, 2022(1): DOI: 10.1155/2022/6255119.
[12]	DONG Jianjun, XIE Zhengquan, JIANG Hao, et al. Multiple regression method for working face mining pressure prediction based on hydraulic support monitoring dataset[J]. Frontiers in Earth Science, 2023, 11:DOI: 10.3389/feart.2023.1114033.
[13]	赵毅鑫, 杨志良, 马斌杰, 等. 基于深度学习的大采高工作面矿压预测分析及模型泛化[J]. 煤炭学报, 2020, 45(1): 54-65.
	ZHAO Yixin, YANG Zhiliang, MA Binjie, et al. Deep learning prediction and model generalization of ground pressure for deep longwall face with large mining height[J]. Journal of China Coal Society, 2020, 45(1): 54-65.
[14]	钱鸣高, 许家林. 煤炭开采与岩层运动[J]. 煤炭学报, 2019, 44(4): 973-984.
	QIAN Minggao, XU Jialin. Behaviors of strata movement in coal mining[J]. Journal of China Coal Society, 2019, 44(4): 973-984.
[15]	秦长坤, 赵武胜, 贾海宾, 等. 基于模态分解和深度学习的煤矿微震时序预测方法[J]. 煤炭学报, 2024, 49(9): 3781-3797.
	QIN Changkun, ZHAO Wusheng, JIA Haibin, et al. A method for predicting the time series of microseismic events in coal mines based on modal decomposition and deep learning[J]. Journal of China Coal Society, 2024, 49(9): 3781-3797.
[16]	韩中合, 朱霄珣, 李文华. 基于K-L散度的EMD虚假分量识别方法研究[J]. 中国电机工程学报, 2012, 32(11): 112-117, 154.
	HAN Zhonghe, ZHU Xiaoxun, LI Wenhua. A false component identification method of EMD based on kullback-leibler divergence[J]. Proceedings of the CSEE, 2012, 32(11): 112-117,154.
[17]	赵润. 智能环境中基于RFID的无源感知关键技术研究[D]. 上海: 上海交通大学, 2019.
	ZHAO Run. Boosting RFID based passive sensing in smart environment[D]. Shanghai: Shanghai Jiaotong University, 2019.
[18]	GAO Tianyu, LI Zhanli, LI Hongan. A prediction network for hydraulic support pressure based on multitimescale feature extraction[J]. Mathematical Problems in Engineering, 2022, 2022(1): DOI: 10.1155/2022/8526744.
[19]	GAO Yifei, ZHANG Xiaohang, LI Zhengren, et al. An attention-based spatio-temporal model for methane concentration forecasting in coal mine[J]. Neural Processing Letters, 2023, 55(4): 4777-4798. doi: 10.1007/s11063-022-11065-4
[20]	HAN Shuzhen, SUN Shengke, ZHAO Zhanshan, et al. Deep residual multiscale convolutional neural network with attention mechanism for bearing fault diagnosis under strong noise environment[J]. IEEE Sensors Journal, 2024, 24(6): 9073-9081. doi: 10.1109/JSEN.2023.3345400
[21]	王红伟, 伍永平, 解盘石, 等. 大倾角变角度综放工作面顶板运移与支架稳定性分析[J]. 中国矿业大学学报, 2017, 46(3): 507-513.
	WANG Hongwei, WU Yongping, XIE Panshi, et al. Research on strata movement and support stability of fully mechanized sublevel caving workface with variable angle in steeply dipping seam[J]. Journal of China University of Mining & Technology, 2017, 46(3): 507-513.
[22]	YU Yong, SI Xiaosheng, HU Changhua, et al. A review of recurrent neural networks: LSTM cells and network architectures[J]. Neural Computation, 2019, 31(7): 1235-1270. doi: 10.1162/neco_a_01199 pmid: 31113301
[23]	康帅, 李治甫, 王自法, 等. 基于相关函数和卷积神经网络的结构损伤识别[J]. 地震工程与工程振动, 2024, 44(2): 50-60.
	KANG Shuai, LI Zhifu, WANG Zifa, et al. Structural damage identification based on correlation function and CNN[J]. Earthquake Engineering and Engineering Dynamics, 2024, 44(2): 50-60.

IMF分量	K-L值	IMF分量	K-L值
IMF 1	0.018	IMF 7	0.269
IMF 2	0.028	IMF 8	0.312
IMF 3	0.009	IMF 9	0.459
IMF 4	0	IMF 10	1
IMF 5	0.030	IMF 11	0.758
IMF 6	0.071	—	—

类型	输入形状	卷积核	通道数	输出形状
Input Layer	(None, 12, 6)	—	12	(None, 12, 6)
Conv1D	(None, 12, 6)	1	16	(None, 12, 16)
Conv1D	(None, 12, 6)	3	16	(None, 12, 16)
Conv1D	(None, 12, 6)	5	16	(None, 12, 16)
Concatenate	(None, 12, 6)	—	48	(None, 12, 48)
TCN+TAM	(None, 12, 48)	3	48	(None, 12, 24)
BatchNormalization	(None, 12, 24)	—	48	(None, 12, 24)
TCN+TAM	(None, 12, 24)	3	48	(None, 12, 12)
LSTM	(None, 12, 12)	—	32	(None, 32)
Dense	(None, 32)	—	5	(None, 5)

试验	隐藏神经元数	RMSE	R²	MAE	T/s
A	16	3.28	0.76	2.46	0.41
B	32	2.48	0.86	1.77	0.44
C	64	2.66	0.84	2.05	0.46
D	128	2.94	0.81	2.03	0.49

方法	4号液压支架			60号液压支架			100号液压支架
方法	RMSE	R²	MAE	RMSE	R²	MAE	RMSE	R²	MAE
MTAM-LSTM	4.32	0.79	3.14	5.37	0.79	4.01	4.38	0.75	3.36
无CEEMDAN	4.53	0.73	3.34	6.08	0.71	4.59	4.62	0.72	3.42
VMD-MTAM-LSTM	4.50	0.71	3.05	6.1	0.74	4.2	4.87	0.70	3.49
LSTM^[13]	4.87	0.73	3.45	7.11	0.64	5.30	4.92	0.69	3.69
BP^[10]	4.48	0.77	3.27	7.50	0.60	5.84	5.56	0.60	4.32

模型	关键模块	RMSE	R²	MAE
MCNN-LSTM	MCNN、LSTM	4.75	0.75	3.31
Model_1	TAM	5.41	0.67	3.94
Model_2	MTAM	4.84	0.74	3.64
LSTM	LSTM	4.87	0.73	3.45