考慮多點(diǎn)監(jiān)測(cè)數(shù)據(jù)的混凝土壩智能預(yù)警分析方法

打開文本圖片集

關(guān)鍵詞：混凝土壩；多點(diǎn)變形監(jiān)測(cè);預(yù)警指標(biāo);K-means聚類法;ConvLSTM模型;3-Sigma原則中圖分類號(hào)：TV62 文獻(xiàn)標(biāo)志碼：Adoi：10.3969/j.issn.1000-1379.2025.07.024引用格式：，李炎隆，張野，等.考慮多點(diǎn)監(jiān)測(cè)數(shù)據(jù)的混凝土壩智能預(yù)警分析方法[J].人民黃河，2025，47（7）：150-155.

Intelligent Early Warning Analysis Method for Concrete Dams Considering Multi-Point Monitoring Data

ZHONG Wen1，LI Yanlong1， ZHANG Ye1， ZHOU Tao2， KANG Xinyu1，YANG Tao3， LI Kangping4 （1.StateKeyLboratoryofWaterEgeringEolodEnviontinAidreaXi'nUivesityfToloXi'na; 2.Huanghe HydropowerDevelopmentCo.，Ltd.，，Xining 810o，China；3.ChinaYangtzePowerCo.，Ltd.，Yibin 644612，China; 4.Power China Northwest Engineering Corporation Limited，Xi’an 71OO65，China）

Abstract：Inodertohancetheaccacyofarlywaringincocretedmsafetymonitoring，isstudypropedanintellgentearlyainganalysismetdbasedonulti-poitmitoringdataimingtoovercoetesuseptibilityoftraditioalsingle-poitmetodstoo structuralinteferec.FirstlyK-eanslusterngmethodasusedtoartiomoitoringontsihsiilaeforationpatesn ConvLSTMmodelwasemployed toextractthespatial-temporalfeaturesoftedformationsequenesfromeachclusterandmakepredictions. Byanalyzingtheresidualsequencesanddeterminingtheearlywarning treshldbasedonthe3-Sigmapriciple，single-pointearlywaing results weregenerated.Finall，teearlywaringresultsfromallusterswereitegratedtoensurethatanearlywaingwastriggdonly whenallmoitorgpoitswitinaustereibitaomalssiultaneoslyattesaetie.Experimentalresultsshowatteproposd methodreducesthefalsealamsandmiseddetectioscausedbyextealdisturbancesinsingle-pointearlywaingmetodsbyintegatig multi-point information，thereby improving the reliabilityand stabilityof the early warning system.

Keywords：concretedam；multi-ointdeforationmonitoring;arlywaingindicators；K-meansclustering method;onSTMmodel;3 Sigma principle

0 引言

大壩作為重要的水利設(shè)施，其安全性直接關(guān)系到人民生命財(cái)產(chǎn)和生態(tài)環(huán)境的安全[1-2]。（剩余7194字）