基于CNN-LSTM模型的露天矿边坡表面变形预测

doi:10.16265/j.cnki.issn1003-3033.2025.S2.0030

中国安全科学学报 ›› 2025, Vol. 35 ›› Issue (S2): 231-238.doi: 10.16265/j.cnki.issn1003-3033.2025.S2.0030

基于CNN-LSTM模型的露天矿边坡表面变形预测

蔡博渊¹^,²(), 于正兴¹^,², 任义¹^,²^,^**(), 张亦海¹^,², 马海涛¹, 王一丹¹^,²

¹ 中国安全生产科学研究院, 北京 100012
² 中安国泰(北京)科技发展有限公司, 北京 102209

收稿日期:2025-08-28 出版日期:2026-02-04
通信作者:
**任义(1992—),男,安徽淮北人,博士,助理研究员,主要从事边坡监测预警、岩石力学、地球物理等方面的工作。E-mail: rymr0611@gmail.com。
作者简介:
蔡博渊 (2000—),男,广东惠州人,硕士,助理工程师,主要从事边坡监测预警、岩土结构抗震等方面的工作。E-mail:caiboyuan0815@163.com。
于正兴正高级工程师
任义助理研究员
张亦海高级工程师
马海涛正高级工程师
基金资助:
应急管理部重点科技计划项目(2024EMST080802); 中国安全生产科学研究院基本科研业务费专项资金资助(2024JBKY20)

Surface deformation prediction of open-pit mine slopes based on CNN-LSTM model

CAI Boyuan¹^,²(), YU Zhengxing¹^,², REN Yi¹^,²^,^**(), ZHANG Yihai¹^,², MA Haitao¹, WANG Yidan¹^,²

¹ China Academy of Safety Science and Technology, Beijing 100012, China
² Cathay Safety Technology Co., Ltd., Beijing 102209, China

Received:2025-08-28 Published:2026-02-04

摘要/Abstract

摘要：

为满足露天矿边坡表面变形高精度预测的工程需求,首先,构建卷积长短时记忆网络(CNN-LSTM)混合预测模型,并以某露天矿为案例,基于边坡合成孔径雷达(S-SAR)获取时序变形数据,采用小波降噪对原始数据进行去噪预处理,提升数据质量;其次,引入自回归积分移动平均模型(ARIMA)作为传统统计方法基准模型,分析基于深度学习模型进行预测的必要性;最后,将CNN-LSTM混合预测模型与单一结构的长短期记忆网络(LSTM)和卷积神经网络(CNN)模型进行对比,验证所构建模型的科学性和有效性。结果表明:LSTM、CNN与CNN-LSTM模型在均方根误差(RMSE)、平均绝对误差(MAE)、决定系数(R²)和平均绝对百分比误差(MAPE)等指标上均优于ARIMA模型,证明边坡变形序列具有显著的非线性特征。在深度学习模型对比中,CNN-LSTM模型表现最为优越。相较于LSTM模型,其RMSE和MAE分别降低55.54%和59.24%;相较于CNN模型,分别降低48.63%和52.15%。此外,该混合模型的R²与MAPE较LSTM模型分别提升5.29%和59.80%,较CNN模型分别提升3.27%和52.51%。

关键词: 露天矿边坡, 变形预测, 卷积长短时记忆网络(CNN-LSTM), 边坡合成孔径雷达(S-SAR), 小波降噪

Abstract:

This study aims to meet the engineering demand for high-precision prediction of surface deformation in open-pit mine slopes. Firstly, a hybrid prediction model of CNN-LSTM was constructed. With a certain open-pit mine taken as a case, time-series deformation data were acquired using S-SAR. Wavelet denoising was applied to preprocess the raw data and enhance data quality. An autoregressive integrated moving average (ARIMA) model was then introduced as the baseline for traditional statistical methods to evaluate the necessity of employing deep learning-based prediction approaches. Finally, the hybrid prediction model of CNN-LSTM was compared with single-structure LSTM and CNN models to verify the scientific soundness and effectiveness of the proposed models. The results show that the LSTM, CNN, and CNN-LSTM models all outperform the ARIMA model in terms of root mean square error (RMSE), mean absolute error (MAE), coefficient of determination (R²), and mean absolute percentage error (MAPE), demonstrating the pronounced nonlinear characteristics of slope-deformation sequences. Among the deep learning architectures, the CNN-LSTM model exhibits the best overall performance. Relative to those of the LSTM model, its RMSE and MAE are reduced by 55.54% and 59.24%, respectively. Relative to those of the CNN model, they are reduced by 48.63% and 52.15%, respectively. In addition, the hybrid model's R² and MAPE are improved by 5.29% and 59.80% compared with those of the LSTM model, and by 3.27% and 52.51% compared with those of the CNN model.

Key words: open-pit mine slope, deformation prediction, convolutional neural network-long short-term memory network (CNN-LSTM), slope-synthetic aperture radar (S-SAR), wavelet denoising

中图分类号:

X915.5

蔡博渊, 于正兴, 任义, 张亦海, 马海涛, 王一丹. 基于CNN-LSTM模型的露天矿边坡表面变形预测[J]. 中国安全科学学报, 2025, 35(S2): 231-238.

CAI Boyuan, YU Zhengxing, REN Yi, ZHANG Yihai, MA Haitao, WANG Yidan. Surface deformation prediction of open-pit mine slopes based on CNN-LSTM model[J]. China Safety Science Journal, 2025, 35(S2): 231-238.

图/表 11

图1

图2

图3

表1

表2

图4

图5

表3

图6

图7

图8

参考文献 13

[1]	GONG Wenping, ZHANG Shaoyan, JUANG H C, et al. Displacement prediction of landslides at slope-scale: review of physics-based and data-driven approaches[J]. Earth-Science Reviews, 2024, 258:DOI:10.1016/j.earscirev.2024.104948.
[2]	ZHOU Xiaoping, LIU Linjiang, XU Ce. A modified inverse-velocity method for predicting the failure time of landslides[J]. Engineering Geology, 2020, 268:DOI:10.1016/j.enggeo.2020.105521.
[3]	WANG Chuang, WANG Zidong, DONG Hongli, et al. Fusionformer: a novel adversarial transformer utilizing fusion attention for multivariate anomaly detection[J]. IEEE Transactions on Neural Networks and Learning Systems, 2025, 36:DOI:10.1109/TNNLS.2025.3542719.
[4]	任义, 于正兴, 张亦海, 等. S-SAR型边坡雷达监测数据去噪及滑坡预报研究[J]. 中国安全生产科学技术, 2025, 21(5): 111-116.
	REN Yi, YU Zhengxing, ZHANG Yihai, et al. Denoising of S-SAR radar monitoring data and landslide forecasting[J]. Journal of Safety Science and Technology, 2025, 21(5): 111-116.
[5]	张亦海, 张怡斌, 马海涛, 等. 降雨条件下含水土质边坡滑坡监测预警试验研究[J]. 中国安全生产科学技术, 2023, 19(增1):5-10.
	ZHANG Yihai, ZHANG Yibin, MA Haitao, et al. Experimental study on landslide monitoring and early warning of water-bearing soil slopes under rainfall conditions[J]. Journal of Safety Science and Technology, 2023, 19(S1): 5-10.
[6]	ZHANG Yihai, MA Haitao, YU Zhengxing. Application of the method for prediction of the failure location and time based on monitoring of a slope using synthetic aperture radar[J]. Environmental Earth Sciences, 2021, 80(21):1-12. doi: 10.1007/s12665-020-09327-2
[7]	YAN Kai, WU Jiaye, ZHANG Yingqian, et al. Short-term forecasting management research of deep horizontal displacement of slope soil based on ARIMA model[J]. Open Access Library Journal, 2019, 6(6): 1-18.
[8]	WANG Zhao, TANG Jiakui, HOU Shengshan, et al. Landslide displacement prediction from on-site deformation data based on time series ARIMA model[J]. Frontiers in Environmental Science, 2023, 11:DOI:10.3389/fenvs.2023.1249743.
[9]	MA Ning, YAO Zaizhen. Analysis of slope stochastic fields using a novel deep learning model with attention mechanism[J]. Frontiers in Earth Science, 2024, 12(12):DOI:10.3389/feart.2024.1352958.
[10]	XI Ning, YANG Qiang, SUN Yingjie, et al. Machine learning approaches for slope deformation prediction based on monitored time-series displacement data: a comparative investigation[J]. Applied Sciences, 2023, 13(8): DOI:10.3390/app13084677.
[11]	WANG Sichan, LYU Tianle, LUO Naqiang, et al. Deformation prediction of rock cut slope based on long short-term memory neural network[J]. International Journal of Machine Learning and Cybernetics, 2024, 15(3):795-805. doi: 10.1007/s13042-023-01939-x
[12]	ZHANG Sherong, JIA He, WANG Chao. Deep-learning-based landslide early warning method for loose deposits slope coupled with groundwater and rainfall monitoring[J]. Computers and Geotechnics, 2024, 165:DOI:10.1016/j.compgeo.2023.10592.4.
[13]	LI Limin, WANG Chaoyang, WEN Zongzhou, et al. Landslide displacement prediction based on ICEEMDAN, ApEn and the CNN-LSTM models[J]. Journal of Mountain Science, 2023, 20(5): 1 220-1 231. doi: 10.1007/s11629-022-7606-0

监测区域	指标	降噪前	降噪后	优化率/ %
JC1	信号方差	9 682.053 5	9 676.281 9	0.06
	RMSE	—	2.469 8	—
	SNR	—	31.99 dB	—
JC2	信号方差	9 526.606 7	9 488.116 7	0.40
	RMSE	—	2.650 9	—
	SNR	—	31.29 dB	—
JC3	信号方差	6 027.168 0	6 013.736 8	0.22
	RMSE	—	1.692 6	—
	SNR	—	33.21 dB	—

模型	RMSE	MAE	R²	MAPE
ARIMA	16.046	11.751	0.659	6.553
LSTM	11.252	9.804	0.956	3.496
CNN	9.335	8.571	0.970	3.224
CNN-LSTM	6.445	5.261	0.986	1.901

监测区域	模型	总训练时间/s	收敛轮次
JC1	LSTM	210	125
	CNN	180	100
	CNN-LSTM	150	100
JC2	LSTM	210	12
	CNN	180	110
	CNN-LSTM	150	100
JC3	LSTM	210	120
	CNN	180	115
	CNN-LSTM	150	105

基于CNN-LSTM模型的露天矿边坡表面变形预测

Surface deformation prediction of open-pit mine slopes based on CNN-LSTM model

RichHTML

PDF

可视化

摘要/Abstract

引用本文

使用本文

图/表 11

参考文献 13

相关文章 4

编辑推荐

Metrics

本文评价

[1]	冯巩, 夏元友, 王智德, 严敏嘉. 基于位移信息融合的露天矿边坡动态预警方法[J]. 中国安全科学学报, 2022, 32(3): 116-122.
[2]	张颖, 张维, 徐磊, 王兵. 大型储罐底板焊缝缺陷非接触式自动检测仪的研制[J]. 中国安全科学学报, 2018, 28(10): 67-72.
[3]	岩小明，李夕兵，陈祥云. 基于距离判别分析理论的露天矿边坡潜在破坏模式识别方法[J]. 中国安全科学学报, 2012, 22(8): 124-.
[4]	陈光辉，李夕兵，董陇军，李地元，刘新颖. 排土场与爆破荷载共同作用下的露天矿边坡稳定性分析[J]. 中国安全科学学报, 2012, 22(5): 113-.