波致海床瞬态液化对再悬浮贡献率的深度学习预测方法

冯春健 刘汉露 刘锦昆 贾永刚 侯方 薛凉 权永峥

冯春健, 刘汉露, 刘锦昆, 等. 2021.波致海床瞬态液化对再悬浮贡献率的深度学习预测方法[J].工程地质学报, 29(6): 1788-1795. doi: 10.13544/j.cnki. jeg.2021-0423
引用本文: 冯春健, 刘汉露, 刘锦昆, 等. 2021.波致海床瞬态液化对再悬浮贡献率的深度学习预测方法[J].工程地质学报, 29(6): 1788-1795. doi: 10.13544/j.cnki. jeg.2021-0423
Feng Chunjian, Liu Hanlu, Liu Jinkun, et al. 2021. Prediction method for contribution rate of wave-induced seabed transient liquefaction to resuspension based on deep learning [J].Journal of Engineering Geology, 29(6): 1788-1795. doi: 10.13544/j.cnki.jeg.2021-0423
Citation: Feng Chunjian, Liu Hanlu, Liu Jinkun, et al. 2021. Prediction method for contribution rate of wave-induced seabed transient liquefaction to resuspension based on deep learning [J].Journal of Engineering Geology, 29(6): 1788-1795. doi: 10.13544/j.cnki.jeg.2021-0423

波致海床瞬态液化对再悬浮贡献率的深度学习预测方法

doi: 10.13544/j.cnki.jeg.2021-0423
基金项目: 

国家自然科学基金 41877223

海底管道灾害性地质(滑坡等)原位监测系统先导试验研究 20200210

详细信息
    作者简介:

    冯春健(1971-),男,硕士,高级工程师,主要从事海洋平台和海底管道的设计研究和项目管理工作. E-mail: fengcj.osec@sinopec.com

    通讯作者:

    刘汉露(1993-),男,博士生,主要从事海洋工程地质方面的科研工作. E-mail: hanluliu1896@163.com

  • 中图分类号: P67

PREDICTION METHOD FOR CONTRIBUTION RATE OF WAVE-INDUCED SEABED TRANSIENT LIQUEFACTION TO RESUSPENSION BASED ON DEEP LEARNING

Funds: 

the National Natural Science Foundation of China 41877223

Research on Guide for In-situ Monitoring Systems for Catastrophic Geology (landslides, etc.) of Submurine Pipelines 20200210

  • 摘要: 波致瞬态液化渗流导致海床内细粒沉积物向海水中运移,这一过程对海底沉积物再悬浮的贡献率不容忽视,但是贡献率的准确估计和预测比较困难。本研究将黄河水下三角洲的观测数据(包括水深、有效波高、有效波周期、实验舱内悬沙浓度、实验舱外悬沙浓度)作为模型输入数据集,基于长短时记忆循环神经网络建立了瞬态液化对再悬浮贡献率的深度学习预测模型。为了客观评价模型的性能,以平均绝对百分比误差、均方根误差和平均平方误差-标准偏差为评判标准,将该深度学习模型与其他预测模型(支持向量回归模型、人工神经网络)的预测结果进行了比较。结果表明,基于长短时记忆循环神经网络的深度学习模型对3.5d以内的瞬态泵送再悬浮贡献率预测误差最小,其平均绝对百分比误差、均方根误差和平均平方误差-标准偏差分别为5.87%、1.6730、0.1574。因此,该模型可以有效地减少机器学习方法在连续预测中产生的误差叠加问题。
  • 图  1  研究区地理位置: (a)现代黄河水下三角洲地理位置(b)观测点位置(c)观测所用到的仪器设备

    Figure  1.  Location of the study area: (a) Geographical location of the underwater delta of the modern Yellow River(b) Position of observation point(c) Instruments and equipment used for observation

    图  2  LSTM的训练和预测数据

    Figure  2.  Training and prediction data of LSTM

    图  3  用于数据预测的长短时记忆循环神经网络结构

    Figure  3.  The structure of short-and long-term memory cyclic neural network for data prediction

    图  4  83.5h内实测与预报贡献率的散点图

    Figure  4.  Scatter plot of the measured and forecast contribution rate within 83.5 hours

    a. SVR; b. ANN; c. LSTM

    图  5  各模型预测结果

    Figure  5.  Prediction results of each model

    a. SVR; b. ANN; c. LSTM

    图  6  各模型不同预测时长结果对比

    Figure  6.  Comparison of the results of different prediction durations of different models

    a. SVR; b. ANN; c. LSTM

    图  7  各模型不同预测时间MAPE的值

    Figure  7.  MAPE values of various models at different prediction times

    图  8  各模型不同预测时间RMSE的值

    Figure  8.  RMSE values of each model at different prediction times

    表  1  SVR,ANN,LSTM模型的MAPE值、RMSE值、RSR

    Table  1.   MAPE values, RMSE values and RSR values of SVR, ANN and LSTM models

    模型 MAPE RMSE RSR
    SVR 39.96 9.8839 0.7526
    ANN 19.64 4.6435 0.3536
    LSTM 5.87 1.6730 0.1574
    下载: 导出CSV
  • Dey S,Ali S Z,Padhi E. 2018. Advances in analytical modeling of suspended sediment transport[J]. Journal of Hydro-environment Research,20 : 110-126. doi: 10.1016/j.jher.2018.02.004
    Du X, Sun Y F, Song Y P, et al. 2020. Multilayer perception neural network for assessment and prediction of earthquake-induced sand liquefaction[J]. Journal of Engineering Geology, 28 (6): 1425-1432.
    Goldstein E B, Coco G, Plant N G. 2019. A review of machine learning applications to coastal sediment transport and morphodynamics[J]. Earth-Science Reviews, 194 : 97-108. doi: 10.1016/j.earscirev.2019.04.022
    Graves A, Jaitly N, Mohamed A R. 2013. Hybrid speech recognition with deep bidirectional LSTM[C]//Automatic speech Recognition and Understanding (ASRU), 2013 IEEE Workshop on IEEE. [S.L.]: [s.n.]
    Green M O, Coco G. 2014. Review of wave-driven sediment resuspension and transport in estuaries[J]. Reviews of Geophysics, 52 (1): 77-117. doi: 10.1002/2013RG000437
    Jeng D S. 2003. Wave-induced sea floor dynamics[J]. Applied Mechanics Reviews, 56(4): 407. doi: 10.1115/1.1577359
    Hochreiter S, Schmidhuber J. 1997. Long short-term memory[J]. Neural Computation, 9 (8): 1735-1780. doi: 10.1162/neco.1997.9.8.1735
    Huang C C, Chang M J, Lin G F, et al. 2021. Real-time forecasting of suspended sediment concentrations reservoirs by the optimal integration of multiple machine learning techniques[J]. Journal of Hydrology: Regional Studies, 34: 100804. doi: 10.1016/j.ejrh.2021.100804
    Kaveh K, Kaveh H, Bui M D, et al. 2020. Long short-term memory for predicting daily suspended sediment concentration[J]. Engineering with Computers, 37 (1): 2013-2027.
    Liu Y H, Fang R K, Sun Y C, et al. 2021. Machine learning based model for warning of regional landslide disasters[J]. Journal of Engineering Geology, 29 (1): 116-124. doi: 10.1088/1755-1315/783/1/012074
    Maa P Y, Kwon J I. 2007. Using ADV for cohesive sediment settling velocity measurements[J]. Estuarine Coastal & Shelf Science, 73(1-2): 351-354. http://www.researchgate.net/profile/Jerome_Maa/publication/228907015_Using_ADV_for_cohesive_sediment_settling_velocity_measurements/links/56254e1b08ae4d9e5c4bb259.pdf
    Maanen B V, Coco G, Bryan K R, et al. 2010. The use of artificial neural networks to analyze and predict alongshore sediment transport[J]. Copernicus Publications, 17 (5): 395-404. http://www.onacademic.com/detail/journal_1000040545937810_f6e5.html
    Mohammad Z K, Amin M M, Meysam A, et al. 2020. On the complexities of sediment load modeling using integrative machine learning: Application of the great river of Loíza in Puerto Rico[J]. Journal of Hydrology(Amsterdam), 585: 124759.
    Mohammad Z K, Özgür K, Jan A, et al. 2016. Evaluation of data driven models for river suspended sediment concentration modeling[J]. Journal of Hydrology, 535 : 457-472. doi: 10.1016/j.jhydrol.2016.02.012
    Rumelhart D E, Hinton G E, Williams R J. 1986. Learning representations by back-propagating errors[J]. Nature, 323(6088): 533-536. doi: 10.1038/323533a0
    Srivastava N, Hinton G, Krizhevsky A, et al. 2014. Dropout: A simple way to prevent neural networks from overfitting[J]. Journal of Machine Learning Research, 15 (1): 1929-1958. http://faculty.dbmi.pitt.edu/day/Bioinf2132-advanced-Bayes-and-R/index.php?path=previousDocuments/Bioinf2132-documents-2016/2016-12-15/&download=JMLRdropout.pdf
    Vapnik V, Golowich S E, Smola A. 1997. Support vector method for function approximation, regression estimation, and signal processing[J]. Advances in Neural Information Processing Systems, 9(2008): 281-287. http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.41.3139&rep=rep1&type=pdf
    Xu X B, Xu G H, Yang J J, et al. 2021. Field observation of the wave-induced pore pressure response in a silty soil seabed[J]. Geo-Marine Letters, 41(1): 13. doi: 10.1007/s00367-020-00680-6
    Xue C. 1993. Historical changes in the Yellow River delta, China[J]. Marine Geology, 113(3-4): 321-330. doi: 10.1016/0025-3227(93)90025-Q
    Yan H, Li S H, Wu L Z. 2019. Landslide displacement prediction based on multiple data-driven model methods[J]. Journal of Engineering Geology, 27 (2): 459-465. http://en.cnki.com.cn/Article_en/CJFDTotal-GCDZ201902028.htm
    Zhang S T, Jia Y G, Wen M Z, et al. 2017. Vertical migration of fine-grained sediments from interior to surface of seabed driven by seepage flows-'sub-bottom sediment pump action'[J]. Journal of Ocean University of China, 16 (1): 15-24. doi: 10.1007/s11802-017-3042-0
    Zhang S T, Jia Y G, Liu X L, et al. 2016. Feature and mechanism of sediment dynamic changing processes in the modern Yellow River delta[J]. Marine Geology and Quaternary Geology, 36 (6): 33-44. http://en.cnki.com.cn/Article_en/CJFDTOTAL-HYDZ201606006.htm
    Zhang S T, Jia Y G, Zhang Y Q, et al. 2018. In situ observations of wave pumping of sediments in the Yellow River Delta with a newly developed benthic chamber[J]. Marine Geophysical Research, 39 (4): 463-474. doi: 10.1007/s11001-018-9344-9
    杜星, 孙永福, 宋玉鹏, 等. 2020. 基于MPL神经网络的地震作用下砂土液化评估及预测[J]. 工程地质学报, 28 (6): 1425-1432. doi: 10.13544/j.cnki.jeg.2019-321
    刘艳辉, 方然可, 苏永超, 等. 2021. 基于机器学习的区域滑坡灾害预警模型研究[J]. 工程地质学报, 29 (1): 116-124. doi: 10.13544/j.cnki.jeg.2020-533
    鄢好, 李绍红, 吴礼舟. 2019. 联合多种数据驱动建模方法的滑坡位移预测研究[J]. 工程地质学报, 27 (2): 459-465. doi: 10.13544/j.cnki.jeg.2017-485
    杨作升. 1993. 埕岛油田勘探开发海洋环境[M]. 青岛: 青岛海洋大学出版社.
    张少同, 贾永刚, 刘晓磊, 等. 2016. 现代黄河三角洲沉积物动态变化过程的特征与机理[J]. 海洋地质与第四纪地质, 36 (6): 33-44. https://www.cnki.com.cn/Article/CJFDTOTAL-HYDZ201606006.htm
  • 加载中
图(8) / 表(1)
计量
  • 文章访问数:  120
  • HTML全文浏览量:  23
  • PDF下载量:  23
  • 被引次数: 0
出版历程
  • 收稿日期:  2021-06-30
  • 修回日期:  2021-10-20
  • 刊出日期:  2021-12-25

目录

    /

    返回文章
    返回