Page 97 - 2023年第54卷第4期
P. 97
[J].Landslides,2013,10(2):203 - 218.
[23] 黄晓虎,雷德鑫,夏俊宝,等.降 雨 诱 发 滑 坡 阶 跃 型 变 形 的 预 测 分 析 及 应 用 [J].岩 土 力 学,2019,40
(9):3585 - 3602.
[24] 覃瀚萱,桂蕾,余玉婷,等.基于滑坡灾害预警分级的应急处置措施[J].地质科技通报,2021,40(4):
187 - 195.
[25] 罗文强,李飞 翱,刘 小 珊,等.多 元 时 间 序 列 分 析 的 滑 坡 演 化 阶 段 划 分 [J].地 球 科 学,2016,41(4):
711 - 717.
[26] 郭科,彭继兵,许强,等.滑坡多点数据融合中的多传感器目标跟踪技术应用[J].岩土力学,2006,27
(3):479 - 481.
[27] CHENZM,YEH Y L,CHEN TC.Assessmentofaregionalflooddisasterindicatorviaanentropyweighting
method [J].NaturalHazardReview,2018,19(2):05018002.
[28] 韦方强,胡凯衡,陈杰.泥石流预报中前期有效降水量的确定[J].山地学报,2005,23(4):453 - 457.
[29] 邓聚龙.灰色系统理论教程[M].武汉:华中理工大学出版社,1990.
[30] LEOB.Random Forests[J].MachineLearning,2001,45(1):5 - 32.
Researchoncomprehensivedeformationpredictionandmonitoringand
earlywarningmethodforstep - likelandslide
3
1,2,3
4
1
4
5
YUANWei ,SUNRuifeng,ZHONGHuiya,JIAOHaiming,HUHuihua,LINHang
(1.PowerChinaZhongnanEngineeringCorporationLimited,Changsha 410014,China;
2.SchoolofCivilEngineering,ShijiazhuangTiedaoUniversity,Shijiazhuang 050043,China;
3.SchoolofResourcesandSafetyEngineering,CentralSouthUniversity,Changsha 410012,China;
4.ChinaRailway17thBureauGroupUrbanConstructionCo.,Ltd.,Guiyang 550029,China;
5.HunanProvincialCommunicationsPlanning,SurveyandDesignInstituteCO.LTD.,Changsha 410200,China)
Abstract:Thelong - termdeformationofthebankslopeofthelargereservoirpresentsanobviousperiodical“step -
type ”steepincreaseundertheactionofrainfallinfloodseason.Aimingatthedeformationcharacteristicsofstep
landslide ,athree - steplandslidemonitoringandearlywarningmethodformulti - sourcedataisproposedinthis
study:(1)“Fusion”-Themulti - pointmonitoringdataaredecomposedintotrendtermandperiodtermbasedon
empiricalmodedecompositionmethod ,thetrendterm andperiodicterm displacementsofdifferentmonitoring
pointsarefusedbyweightingmethod ,andtheobtainedtrendtermandperiodictermaresuperimposedtoobtainthe
currentstatusoftheintegratedlandslidedeformation.(2)“Prediction”-Theconceptof “onepredictionperiod”
isintroduced,theslidingpolynomialfittingmethodandtherandom forestalgorithm areusedtopredictthefusion
trendtermandthefusionperiodtermrespectivelyandsuperimposethemtoobtainthepredictedcomprehensivede
formationofthelandslide.(3)“Earlywarning”-Basedontheslopechangepointtheory ,the“stablepoint”and
“transitionpoint”ofthecomprehensivedeformationcurvearesearched,andtheaverageslopeofthestabledeform
ationandaccelerateddeformationintervalisdetermined.Then ,a4 - stageprogressivehierarchicalwarningmodel
ofstep - likelandslideisestablished ,andthecurrentsafetyofthebankslopeiswarned.Takingtheslopeautomatic
displacementmonitoringdataofabankslopeofXiangjiabaReservoirasanexample,thecomprehensivedeforma
tionpredictionandsafetywarningofthisbankslopearecarriedoutbyusingthemethodproposedinthispaper.The
resultsshowthatthecomprehensivedeformationtimeseriescanreflectthedeformationevolutionlawofthelandslide
asawhole ,andthepredictionresultsarereliable.Accordingtothehierarchicalwarningmodel,thisslopeiscur
rentlyinthestageofstabledeformation(GradeIearly - warning).
Keywords:step - likelandslide;deformationprediction;monitoringandearlywarning;datafusion;empirical
modedecompositionmethod;graycorrelationtheory;randomforestalgorithm;changepointanalysismethod
(责任编辑:李 娜)
— 4 7 3 —