Displacement Prediction of Step-like Landslide Considering Time Series: A Case Study of Muping Landslide in the Three Gorges Reservoir Area
-
摘要:
本文针对三峡库区呈现阶跃式变形特征的滑坡提出一种考虑时间序列的滑坡位移预测模型，该模型首先采用互补集合经验模态分解方法（Complementary Ensemble Empirical Mode Decomposition, CEEMD）将滑坡的累计位移分解为趋势项和波动项，然后采用回归模型拟合趋势项位移；基于滑坡变形特征与诱发因素的响应分析，采用长短时记忆神经网络（Long Short-Term Memory, LSTM）模型进行波动项位移预测；最后将各分项预测位移叠加，实现滑坡累计位移的预测。以三峡库区阶跃型滑坡——墓坪滑坡为例，本文采用CEEMD-LSTM模型进行位移预测，并与卷积神经网络（Convolutional Neural Networks, CNN）、随机森林（Random Forest, RF）和改进的粒子群优化-支持向量机 （Particle Swarm Optimization-Support Vector Machine, PSO-SVM）等模型的预测结果进行对比分析。结果表明，与其他常见的机器学习与深度学习模型相比，考虑时序的CEEMD-LSTM模型的预测精度较高，且在阶跃点的预测优势较为突出。该模型可为实现三峡库区阶跃型滑坡位移实时预测预报提供理论依据和数据支撑。
Abstract:A landslide displacement prediction model considering time series analysis is proposed for the landslides with step-like deformation in the China Three Gorges Reservoir area. The model first employs the Complementary Ensemble Empirical Mode Decomposition (CEEMD) method to decompose the cumulative landslide displacement into trend and fluctuation components. A regression model is then used to fit the trend displacement. Based on the response analysis between landslide deformation characteristics and triggering factors, a Long Short-Term Memory (LSTM) model is adopted to predict the fluctuation displacement. Finally, the predicted displacements of each component are superimposed to achieve cumulative landslide displacement prediction. In this paper, the CEEMD-LSTM model is applied to predict the displacement of Muping landslide in the Three Gorges Reservoir area. Taking the step-like the Muping landslide in the Three Gorges Reservoir area as an example, this study applies the CEEMD-LSTM model for displacement prediction and conducts a comparative analysis with the prediction results of other models, including Convolutional Neural Networks (CNN); Random Forest (RF) and the improved Particle Swarm Optimization-Support Vector Machine (PSO-SVM) models. The results demonstrate that compared to other common machine learning and deep learning models, the time series-aware CEEMD-LSTM model achieves higher prediction accuracy and more prominent prediction advantages at step points. This model could provide theoretical foundations and data support for real-time displacement prediction and early warning of a step-like landslide in the Three Gorges Reservoir area.
-
-
表 1 墓坪滑坡不同阶跃变形阶段的位移、降雨、库水位指标统计表
Table 1. Statistical table of displacement, rainfall, and reservoir water level indicators at the step-like deformation stage of the Muping landslide monitoring points
编号 起始时间
（年/月/日）结束时间
（年/月/日）位移增量
（mm）累积降雨
（mm）日最大降雨
（mm）运行库水位
（m）库水位平均
变化速率（m/d）a 2021/5/10 2021/5/22 277.8 46.7 18.9 157.76~152.35 −0.42 b 2021/6/4 2021/6/23 248.6 57.3 21.8 148.77~146.57 −0.11 c 2021/8/9 2021/8/31 198.5 154 41.6 146.17~157.14 0.5 d 2022/3/20 2022/3/29 256.6 84.8 45.1 165.28~165.39 0.11 e 2022/5/15 2022/6/3 520.9 22.9 8.5 164.04~147.41 −0.83 
下载: 导出CSV
表 2 墓坪滑坡波动项位移预测精度对比
Table 2. The precision comparison of periodic displacement prediction for Muping landslide
评价指标 预测模型 LSTM模型 CNN模型 RF模型 PSO-SVM模型 RMSE 2.139 6.163 2.514 4.241 MAE 1.056 3.254 1.474 1.592 MBE 0.162 −1.386 −0.231 −0.281
下载: 导出CSV
-
[1] 邓冬梅,梁 烨,王亮清,王昌硕,孙自豪,王 聪,董曼曼.2017.基于集合经验模态分解与支持向量机回归的位移预测方法:以三峡库区滑坡为例[J]. 岩土力学,38(12):3660-3669.
[2] 甘柳燕,唐国强,蒋文希,覃良文.2024.基于CEEMD-LSTM-Adaboost模型的白糖期货跨期套利策略[J]. 桂林理工大学学报,44(1):162-167. doi: 10.3969/j.issn.1674-9057.2024.01.023
[3] 高晨曦,石长柏,闫 巍,侯时平,唐 玄,安知利,路永强.2024.堆积层滑坡在库水位变动与降雨共同作用下的阈值研究——以墓坪滑坡为例[J]. 资源环境与工程,38(1):73-82.
[4] 高 睿,张真弼,李立辰,关 鹏.2024.基于BO-SVM的滑坡稳定性可解释机器学习预测方法[J]. 资源环境与工程,38(3):274-284+307.
[5] 李长冬,龙晶晶,姜茜慧,付智勇.2020.水库滑坡成因机制研究进展与展望[J]. 地质科技通报,39(1):67-77.
[6] 李麟玮,吴益平,苗发盛,廖 康,张龙飞.2018.基于变分模态分解与GWO-MIC-SVR模型的滑坡位移预测研究[J]. 岩石力学与工程学报,37(6):1395-1406.
[7] 柳 青,易 武.2019.基于诱发因素响应与BP神经网络的滑坡位移预测预报[J]. 三峡大学学报(自然科学版),41(3):41-45.
[8] 刘艺梁,殷坤龙,汪 洋,王 伟.2013.基于经验模态分解和神经网络的滑坡变形预测研究[J]. 安全与环境工程,20(4):14-17. doi: 10.3969/j.issn.1671-1556.2013.04.004
[9] 卢书强,易庆林,易 武,张国栋,何 祥.2014.库水下降作用下滑坡动态变形机理分析——以三峡库区白水河滑坡为例[J]. 工程地质学报,22(5):869-875.
[10] 马社祥,刘贵忠,曾召华.2000.基于小波分析的非平稳时间序列分析与预测[J]. 系统工程学报,(4):305-311. doi: 10.3969/j.issn.1000-5781.2000.04.001
[11] 徐 峰,汪 洋,杜 娟,叶 疆.2011.基于时间序列分析的滑坡位移预测模型研究[J]. 岩石力学与工程学报,30(4):746-751.
[12] 姚 琦,牛瑞卿,赵金童,张 淼.2017.基于经验模态分解-支持向量机的滑坡位移预测方法研究[J]. 安全与环境工程,24(1):26-32.
[13] 杨背背,殷坤龙,杜 娟.2018.基于时间序列与长短时记忆网络的滑坡位移动态预测模型[J]. 岩石力学与工程学报,37(10):2334-2343.
[14] 张 俊,殷坤龙,王佳佳,黄发明.2015.基于时间序列与PSO-SVR耦合模型的白水河滑坡位移预测研究[J]. 岩石力学与工程学报,34(2):382-391.
[15] 周 浩,朱平华,蒋宏伟,俞宏艳,沈心怡.2024.逻辑回归优化的静-动态耦合模型在滑坡位移预测中的应用[J]. 资源环境与工程,38(4):446-456.
[16] Li C D, Long J J, Liu Y, Huang D W. 2024. Landslide Prediction Model Based Upon Intelligent Processing of Multi-Point Monitoring Information: A Review[M]. //Abolmasov B, et al. Progress in Landslide Research and Technology. Springer, Cham.
[17] Lian C, Zeng Z G, Yao W, Tang H M. 2012. Displacement prediction model of landslide based on a modified ensemble empirical mode decomposition and extreme learning machine[J]. Natural Hazards, 66(2): 759-771.
[18] Long J J, Li C D, Liu Y, Feng P F, Zuo Q J. 2022. A multi-feature fusion transfer learning method for displacement prediction of rainfall reservoir-induced landslide with step-like deformation characteristics[J]. Engineering Geology, 297: 106494.
[19] Huang F M, Yin K L, Zhang G R, Gui L, Yang B B, Liu L. 2016. Landslide displacement prediction using discrete wavelet transform and extreme learning machine based on chaos theory[J]. Environmental Earth Sciences, 75: 1376. doi: 10.1007/s12665-016-6133-0
[20] Huang N E, Shen Z, Long S R, Wu M C, Shih H H, Zheng Q, Yen N C, Tung C C, Liu H H. 1998. The empirical mode decomposition and the Hilbert spectrum for nonlinear and non-stationary time series analysis[J]. Proceedings of the Royal Society: Mathematical, physical and engineering sciences, 454(1971): 903-995.
[21] Hochreiter S, Schmidhuber J. 1997. Unsupervised coding with loco code[C]. Artificial Neural Networks - ICANN '97, 7th International Conference, Lausanne, Switzerland, October 8-10.
[22] Wu Z, Huang N E. 2009. Ensemble Empirical Mode Decomposition: A Noise-saaisted Data Analysis Method[J]. World Scientific, 1: 1-41.
[23] Yeh J R, Shieh J S, Huang N E. 2010. Complementary Ensemble Empirical Mode Decomposition: a Novel Noise Enhanced Data Analysis Method[J]. Advances in Adaptive Data Analysis, 2(2): 135-156. doi: 10.1142/S1793536910000422
[24] Yang B B, Yin K L, Lacasse S, Liu Z Q. 2019. Time series analysis and long short-term memory neural network to predict landslide displacement[J]. Landslides, 16(4): 677-694. doi: 10.1007/s10346-018-01127-x
[25] Zhou C, Yin K L, Cao Y, Intrieri E, Ahmed B, Catani F. 2018. Displacement prediction of step-like landslide by applying a novel kernel extreme learning machine method[J]. Landslides, 15(11): 2211-2225. doi: 10.1007/s10346-018-1022-0
[26] Zhu X, Ma S Q, Xu Q, Xu Q, Liu W D. 2018. A WD-GA-LSSVM model for rainfall-triggered landslide displacement prediction[J]. Journal of Mountain Science, 15(1): 156-166. doi: 10.1007/s11629-016-4245-3
-
图(8)
表(2)
计量
- 文章访问数: 43
- PDF下载数: 12
- 施引文献: 0