Page 133 - 2023年第54卷第6期
P. 133
[19] 靳聪聪,迟世春,聂章博.基于地震变形易损性的高土石坝抗震安全分析 [J].岩土工程学报,2020,42
(2):334 - 343.
[20] 郭秀吉,陈立,颜小飞,等.2008年以来紫坪铺水库淤积过程及库容变化特性分析[J].中国农村水利水
电,2022(7):55 - 60,66.
[21] 杨正权,刘小生,周国斌,等.高面板堆石坝面板地震动力反应特性大型振动台模型试验研究[J].振动工
程学报,2018,31(2):209 - 218.
[22] 刘晶波,宝鑫,李述涛,等.采用黏弹性人工边界时显式算法稳定性条件[J].爆炸与冲击,2022,42(3):
124 - 137.
[23] 孙陶,高希章,杨建.紫坪铺混凝土面板堆石坝应力- 应变分析[J].岩土力学,2006(2):247 - 251.
[24] 楚金旺,朱晟,黄亚梅.基于实际震害的土石坝永久变形估算[J].中国水利水电科学研究院学报,2017,
15(6):409 - 417.
[25] GONGJ,ZOUDG,KONGXJ,etal.Thesimulationofhighcompressivestressandextrusionphenomenonfor
concretefaceslabsinCFRDsunderstrongseismicloads [J].SoilDynamicsandEarthquakeEngineering,2021,
147:106792.
[26] 孔宪京,周晨光,邹德高,等.高土石坝- 地基动力相互作用的影响研究 [J].水利学报,2019,50(12):
1417 - 1432.
Studyofseismicwaveinputofdam bedrockbasedonmodal
decompositionandcloudparticlenetwork
1,3
1,3
1,3
1
1
1,2
ZHANGHongyang ,LITong,YANGYige,DINGZelin ,ZHANGXianqi ,WANGShunsheng
(1.SchoolofWaterConservancy,NorthChinaUniversityofWaterResourcesandElectricPower,Zhengzhou 450046,China;
2.QingYuanCollege,NorthChinaUniversityofWaterResourcesandElectricPower,Xinyang 464200,China;
3.CollaborativeInnovationCenterforEfficientUtilizationofWaterResourcesinYellowRiverBasin,Zhengzhou 450046,China)
Abstract:Theinputofbedrockseismicwaveintheexistingseismicevaluationofdamsmostlyusesactualdataor
manualgeneration,anditbecomesparticularlydifficulttodeterminebedrockseismicwavewhenthedam sitein
strumentationisdamagedorhistoricalinformationisinadequate.Therefore , theresearchideaofinversionof
seismicwaveinputtobedrockofearthandrockdamsisproposed ,andahybriddecomposition - training - inversion
modelbasedonempiricalmodaldecompositionandcloudparticlenetworkwasdeveloped,determinationofseismic
waveofdambedrockbyusingonlyasmallnumberofperipheralmeasurementstationswithoutrelyingonhistorical
seismicdataofthesite.Firstly ,themeasuredseismicwaverecordsofsurfaceandbedrockwereselected,andthe
accelerationsequencewasdecomposedbytheempiricalmodaldecompositionmethod.Secondly ,theparticleswarm
algorithmwasusedtoestablishthemappingwiththeneuralnetworkconnectionweights ,optimizingtheglobal
searchcapabilityofparticleswarmalgorithmsusingcloudtheory ,establishtheinversionmodel,andusethede
composedaccelerationsequenceasthetrainingsetforinversiontraining.Then ,theseismicwaveinformationmeas
uredatthesurface ,whichwasinasimilargeologicalsituationasthedam,wasselectedandcombinedwiththein
versionmodeltoinverttheseismicwaveinputtothebedrockofthedam.Finally ,theZipingpudamisusedasare
searchexampletoverifytheapplicabilityofthemodelbycomparingthetraditionalinputmethods.Theresultsshow
thatthehybridmodelproposedinthispaperhasacomprehensiveandstableperformanceandcaninverttheseismic
accelerationsequencesbetter ,withthemodelcoefficientofdeterminationaregreaterthan0.9,meanabsoluteper
centageerrorareabout11%.Atthesametime ,thecalculationofbedrockseismicwaveobtainedfromtheinversion
ofthispaperreducesthecalculationerrorby0.79%~17.28% comparedwiththeexistingresearchresults ,whichis
moreconsistentwiththeactualdynamicresponseoftheproject,thusprovidinganewwaytosolvetheproblem of
acquiringseismicwavefromthebedrockinputofthedam.
Keywords:seismicwaveinput;earthrockdam;inversion;radialbasisfunctionneuralnetwork;cloudtheory;
particleswarmoptimization;empiricalmodaldecomposition
(责任编辑:韩 昆)
— 7 6 1 —