Please wait a minute...
浙江大学学报(工学版)  2021, Vol. 55 Issue (9): 1705-1713    DOI: 10.3785/j.issn.1008-973X.2021.09.012
土木工程、水利工程     
基于半监督机器学习的滑坡易发性预测建模
黄发明1(),潘李含1,姚池1,周创兵2,姜清辉2,常志璐1
1. 南昌大学 建筑工程学院,江西 南昌 330031
2. 武汉大学 土木建筑工程学院,湖北 武汉 430072
Landslide susceptibility prediction modelling based on semi-supervised machine learning
Fa-ming HUANG1(),Li-han PAN1,Chi YAO1,Chuang-bing ZHOU2,Qing-hui JIANG2,Zhi-lu CHANG1
1. School of Civil Engineering and Architecture, Nanchang University, Nanchang 330031, China
2. School of Civil Engineering, Wuhan University, Wuhan 430072, China
 全文: PDF(947 KB)   HTML
摘要:

为了克服滑坡编录样本不足、扩充滑坡样本较困难、主观随机选择的非滑坡样本准确性较低等缺点,以江西省南康区为例,拟用半监督卡方自交互侦测决策树(SSCHAID)和半监督反向传播神经网络(SSBPNN)进行滑坡易发性预测(LSP), 在已知滑坡样本和随机选取的非滑坡样本基础上,用全监督机器学习将初始LSP划分成不同级别;将高分辨率遥感影像和初始滑坡易发性图中的极高易发区叠加,筛选一定数量的潜在滑坡栅格单元扩充滑坡样本;从极低易发区选取非滑坡栅格单元组合成新的输出变量;将新的输出变量导入全监督机器学习,获得最终LSP并评价其精度. 结果表明:半监督机器学习的LSP精度远高于全监督机器学习的LSP精度.

关键词: 滑坡易发性预测 (LSP)半监督机器学习卡方自交互侦测决策树 (CHAID)BP神经网络(BPNN)地理信息系统(GIS)    
Abstract:

A semi-supervised Chi-square self-interactive detection decision tree (SSCHAID) and a semi-supervised back-propagation neural network (SSBPNN) were used for landslide susceptibility prediction (LSP) by taking the Nankang of Jiangxi Province as a case, in order to overcome the shortcomings such as insufficient landslide inventories, difficulty in expanding landslide inventories and subjectively randomly selected non-landslides have low accuracy. Based on the known landslides and randomly selected non-landslides, the initial LSP was divided into different levels by supervised machine learning. The high-resolution remote sensing image was superimposed with the very high susceptibility area in the initial landslide susceptibility map, and a certain number of potential landslide grids were selected to expand landslide inventories. Non-landslide grids were selected from very low susceptibility areas and combined into new output variables. The new output variables were imported into supervised machine learning to obtain the final LSP and evaluate its accuracy. Results show that the accuracy of LSP by semi-supervised machine learning is significantly higher than that of supervised machine learning.

Key words: landslide susceptibility prediction (LSP)    semi-supervised machine learning    Chi-squared automatic interaction detector (CHAID)    BP neural network (BPNN)    geographic information system (GIS)
收稿日期: 2020-09-04 出版日期: 2021-10-20
CLC:  P 642.22  
基金资助: 国家自然科学基金资助项目(41807285,41762020,51879127,51769014);江西省自然科学基金资助项目(20192BAB216034,20192ACB2102,20192ACB20020);中国博士后面上基金资助项目(2019M652287,2020T130274);江西省博士后基金资助项目(2019KY08);研究生创新专项资金资助项目(YC2020-S120)
作者简介: 黄发明(1988—),男,副教授,博士,从事滑坡易发性预测研究. orcid. org/0000-0001-9037-9085. E-mail: faminghuang@ncu.edu.cn
服务  
把本文推荐给朋友
加入引用管理器
E-mail Alert
作者相关文章  
黄发明
潘李含
姚池
周创兵
姜清辉
常志璐

引用本文:

黄发明,潘李含,姚池,周创兵,姜清辉,常志璐. 基于半监督机器学习的滑坡易发性预测建模[J]. 浙江大学学报(工学版), 2021, 55(9): 1705-1713.

Fa-ming HUANG,Li-han PAN,Chi YAO,Chuang-bing ZHOU,Qing-hui JIANG,Zhi-lu CHANG. Landslide susceptibility prediction modelling based on semi-supervised machine learning. Journal of ZheJiang University (Engineering Science), 2021, 55(9): 1705-1713.

链接本文:

https://www.zjujournals.com/eng/CN/10.3785/j.issn.1008-973X.2021.09.012        https://www.zjujournals.com/eng/CN/Y2021/V55/I9/1705

图 1  半监督机器学习预测滑坡易发性的建模流程
图 2  遥感解译流程
内部控制因素 属性区间 因素类型 主区栅格 滑坡区栅格 $ {\rm{F}}{{\rm{R}}_{{i}}} $
数量 占比/% 数量 占比/%
坡度/ (°) 0~2.95 连续型 570 276 27.64 35 1.347 0.049
坡度/ (°) 2.95~6.27 连续型 465 255 22.55 369 14.203 0.629
坡度/ (°) 6.27~9.78 连续型 342 869 16.62 639 24.596 1.480
坡度/ (°) 9.78~13.28 连续型 276 438 13.40 731 28.137 2.100
坡度/ (°) 13.28~16.97 连续型 200 695 9.73 525 20.208 2.077
坡度/ (°) 16.97~21.21 连续型 125 840 6.10 248 9.546 1.565
坡度/ (°) 21.21~26.93 连续型 62 731 3.04 50 1.925 0.633
坡度/ (°) 26.93~47.03 连续型 18 898 0.92 1 0.038 0.041
距离水系的距离/m >750 离散型 843 343 40.88 347 13.356 0.327
距离水系的距离/m 500~750 离散型 358 995 17.40 277 10.662 0.613
距离水系的距离/m 250~500 离散型 409 427 19.85 619 23.826 1.201
距离水系的距离/m 0~250 离散型 451 237 21.87 1 355 52.156 2.384
地层岩性 变质岩 离散型 815 922 39.55 1 453 55.928 1.414
地层岩性 碳酸盐岩 离散型 688 947 33.40 336 12.933 0.387
地层岩性 碎屑岩 离散型 546 081 26.47 809 31.139 1.176
地层岩性 水域 离散型 12 052 0.58 0.000 0.000 0.000
表 1  滑坡内部控制因素的频率比值
图 3  南康区相关高分辨率航片及部分控制因素
图 4  各模型预测的滑坡易发性分布图
图 5  全监督和半监督机器学习预测滑坡易发性ROC曲线
模型 S PPR TA
CHAID 75.21 74.67 76.65
SSCHAID 79.23 85.04 82.67
BPNN 81.52 76.16 79.58
SSBPNN 88.00 90.55 90.46
表 2  4个模型的统计学精度
1 BUI D T, PRADHAN B, LOFMAN O, et al Landslide susceptibility mapping at Hoa Binh province (Vietnam) using an adaptive neuro-fuzzy inference system and GIS[J]. Computers and Geosciences, 2012, 45: 199- 211
doi: 10.1016/j.cageo.2011.10.031
2 刘渊博, 牛瑞卿, 于宪煜, 等 旋转森林模型在滑坡易发性评价中的应用研究[J]. 武汉大学学报:信息科学版, 2018, 43 (6): 959- 964
LIU Yuan-bo, NIU Rui-qing, YU Xian-yu, et al Application of the rotation forest model in landslide susceptibility assessment[J]. Geomatics and Information Science of Wuhan University, 2018, 43 (6): 959- 964
3 WANG Q, WANG Y, NIU R, et al Integration of information theory, K-means cluster analysis and the logistic regression model for landslide susceptibility mapping in the Three Gorges Area, China[J]. Remote Sensing, 2017, 9 (9): 938
doi: 10.3390/rs9090938
4 黄发明, 殷坤龙, 蒋水华, 等 基于聚类分析和支持向量机的滑坡易发性评价[J]. 岩石力学与工程学报, 2018, 37 (1): 156- 167
HUANG Fa-ming, YIN Kun-long, JIANG Shui-hua, et al Landslide susceptibility assessment based on clustering analysis and support vector machine[J]. Chinese Journal of Rock Mechanics and Engineering, 2018, 37 (1): 156- 167
5 BUI D T, TUAN T A, KLEMPE H, et al Spatial prediction models for shallow landslide hazards: a comparative assessment of the efficacy of support vector machines, artificial neural networks, kernel logistic regression, and logistic model tree[J]. Landslides, 2016, 13 (2): 361- 378
doi: 10.1007/s10346-015-0557-6
6 黄发明, 殷坤龙, 张桂荣, 等 多变量PSO-SVM模型预测滑坡地下水位[J]. 浙江大学学报:工学版, 2015, 49 (6): 1193- 1200
HUANG Fa-ming, YIN Kun-long, ZHANG Gui-rong, et al Prediction of groundwater level in landslide using multivariable PSO-SVM model[J]. Journal of Zhejiang University:Engineering Science, 2015, 49 (6): 1193- 1200
7 HUANG F, ZHANG J, ZHOU C, et al A deep learning algorithm using a fully connected sparse autoencoder neural network for landslide susceptibility prediction[J]. Landslides, 2020, 17 (1): 217- 229
doi: 10.1007/s10346-019-01274-9
8 CHANG Z, DU Z, ZHANG F, et al Landslide susceptibility prediction based on remote sensing images and GIS: comparisons of supervised and unsupervised machine learning models[J]. Remote Sensing, 2020, 12 (3): 502
doi: 10.3390/rs12030502
9 冯杭建, 周爱国, 俞剑君, 等 浙西梅雨滑坡易发性评价模型对比[J]. 地球科学, 2016, 41 (3): 403- 415
FENG Hang-jian, ZHOU Ai-guo, YU Jian-jun, et al A comparative study on plum-rain-triggered landslide susceptibility assessment models in west Zhejiang Province[J]. Earth Science, 2016, 41 (3): 403- 415
10 柯懂湘, 潘丽敏, 罗森林, 等 基于随机森林算法的Android恶意行为识别与分类方法[J]. 浙江大学学报:工学版, 2019, 53 (10): 2013- 2023
KE Dong-xiang, PAN Li-min, LUO Sen-lin, et al Android malicious behavior recognition and classification method based on random forest algorithm[J]. Journal of Zhejiang University: Engineering Science, 2019, 53 (10): 2013- 2023
doi: 10.3785/j.issn.1008-973X.2019.10.019
11 PRADHAN B A comparative study on the predictive ability of the decision tree, support vector machine and neuro-fuzzy models in landslide susceptibility mapping using GIS[J]. Computers and Geosciences, 2013, 51: 350- 365
doi: 10.1016/j.cageo.2012.08.023
12 郭子正, 殷坤龙, 付圣, 等 基于GIS与WOE-BP模型的滑坡易发性评价[J]. 地球科学, 2019, 44 (12): 4299- 4312
GUO Zi-zheng, YIN Kun-long, FU Sheng, et al Evaluation of landslide susceptibility based on GIS and WOE-BP model[J]. Earth Science, 2019, 44 (12): 4299- 4312
13 HUANG F, CAO Z, GUO J, et al Comparisons of heuristic, general statistical and machine learning models for landslide susceptibility prediction and mapping[J]. Catena, 2020, 191: 104580
doi: 10.1016/j.catena.2020.104580
14 胡云青, 邱清盈, 余秀, 等 基于改进三体训练法的半监督专利文本分类方法[J]. 浙江大学学报:工学版, 2020, 54 (2): 331- 339
HU Yu-qing, QIU Qing-ying, YU Xiu, et al Semi-supervised patent text classification method based on improved Tri-training algorithm[J]. Journal of Zhejiang University: Engineering Science, 2020, 54 (2): 331- 339
15 张为, 李远耀, 张泰丽, 等 基于孕灾敏感性分析的高植被覆盖区滑坡地质灾害遥感解译[J]. 安全与环境工程, 2019, 26 (3): 28- 35
ZHANG Wei, LI Yuan-yao, ZHANG Tai-li, et al Remote sensing interpretation of landslide geological hazards in high vegetation coverage area based on hazard sensitivity analysis[J]. Safety and Environmental Engineering, 2019, 26 (3): 28- 35
16 黄发明, 叶舟, 姚池, 等 滑坡易发性预测不确定性: 环境因子不同属性区间划分和不同数据驱动模型的影响[J]. 地球科学, 2020, 45 (12): 4535- 4549
HUANG Fa-ming, YE Zhou, YAO Chi, et al Uncertainties of landslide susceptibility prediction: different attribute interval divisions of environmental factors and different data-based models[J]. Earth Science, 2020, 45 (12): 4535- 4549
17 HONG H, PRADHAN B, XU C, et al Spatial prediction of landslide hazard at the Yihuang area (China) using two-class kernel logistic regression, alternating decision tree and support vector machines[J]. Catena, 2015, 133: 266- 281
doi: 10.1016/j.catena.2015.05.019
18 HUANG F, YIN K, HUANG J, et al Landslide susceptibility mapping based on self-organizing-map network and extreme learning machine[J]. Engineering Geology, 2017, 223: 11- 22
doi: 10.1016/j.enggeo.2017.04.013
19 ZHU L, HUANG L, FAN L, et al Landslide susceptibility prediction modeling based on remote sensing and a novel deep learning algorithm of a cascade-parallel recurrent neural network[J]. Sensors (Basel, Switzerland), 2020, 20 (6): 1576
doi: 10.3390/s20061576
20 冯杭建, 周爱国, 唐小明, 等 中国东南地区隐性滑坡遥感识别研究[J]. 地质论评, 2014, 60 (6): 1370- 1380
FENG Hang-jian, ZHOU Ai-guo, TANG Xiao-ming, et al Study on remote sensing interpretation for implicit landslide in southeast China[J]. Geological Review, 2014, 60 (6): 1370- 1380
[1] 彭令,牛瑞卿,吴婷. 时间序列分析与支持向量机的滑坡位移预测[J]. J4, 2013, 47(9): 1672-1679.
[2] 赵权利, 孙红月, 尚岳全, 王智磊. 承压水孔压的时空变化对边坡稳定性影响[J]. J4, 2013, 47(8): 1366-1372.