結(jié)構(gòu)損傷識(shí)別方法的研究

1、同濟(jì)大學(xué)碩士學(xué)位論文結(jié)構(gòu)損傷識(shí)別方法的研究姓名：柯朝輝申請(qǐng)學(xué)位級(jí)別：碩士專業(yè)：結(jié)構(gòu)工程指導(dǎo)教師：薛松濤20041201奠墅堡璺叢曼～一一曼莖查蘭妻量矍蘭蘭墊鯊奎ABSTRACTInthisdissertation，thestate‘ofartofresearchesaboutstructuredamageidentificationisdepicted。thebackgroundoftheresearchprojectandthecon

2、tentofthisdissertationareoutiinedaIsoGeneticalgorithm，whichimitatesthebiologicalevolutionismofgeneticschoosingandnatufalsel。ctioninbiologyis鋤overallmultidimensionalsearchingmethodbasedonnat啪lselectionprincipleandnaturalg

3、eneticmechanismGeneticalgorithmisusedtooptimizethep越a嫩e地rSOf出edamageds㈣。耋u攤inthisdissertation，andimprovedgeneticOperators，whichareadaptivelychangedwithtbemagnitudeandthegenatictime，areusedwhichCanmorequicklygettheove耀11o

4、ptmizingresultsrSomenumericalsimulationexamples，adoptingfloatcode，appytoafive氆oorflamewhichindicatethatgeneticalgorithm，adoptingimprovedgeneticoperators，wⅢ№moresuitablefordetectingthepositionanddegreeofthedamagedslme孫≈Th

5、eneur越networksareabletoapproximatearbitrarynon1inearmappingrelationshi口sufficientlysoitisapowerfuldamageidentificationtoolforunknownsystems。Ch∞s搬妞p8掛emsrepresentingthecharacteristicsofthestructure，whicha≈tobeusedastheinp

6、uttoneu描n。撕07ks，isoneofthemostimportantsubjectsinthisapproachA粥wparam囂鈀矗黼辯《穗erelativerootm。ansquareellor(RPMS)，isfirstlyproposedbytimehistoryanalysis，whichistobe”5。d矗stheinputvectorstoneuralnetworksSomenumericalsimulatio

7、nexamplesindicatethatRRMShasnothingtOthetypeoftheinpuningseismicwave，whichcan∽alizetbattheneural“。細(xì)orkcandetectthedamageonline，andhasthehighersensitivitythanmodeshapesandfrequenciesTheanalysisresultsshowthatRILMS，usedast

8、heinputvectorstoneu精1netw。蕊hasgoodresul【s沁detectthedamagepositionandthedamagedegreeofthe5tructureAtlastthe恥ngledamageidentificationmethodisrecommendedinthelarge。辯alesus岱ensi。n齲dgeAkinadaBridgeinthisdissertationThestructu

9、reformofsuspensionbridgeise。mplicated，8nd8“8lysis”orkisrelativelycomplexInthisdissertation，structureform，load。h撇c栳“stieand。odelingprocessinthefiniteelementprocedureANSYSareelaboratelyexpoundedatfirstandfreq“8∽1。sof￡hesus