资源描述：
《关于混沌理论和支持向量回归的铀资源价格预测》由会员上传分享，免费在线阅读，更多相关内容在学术论文-天天文库。
1、东华理工大学硕士学位论文II万方数据Abstract东华理工大学研究生毕业论文英文摘要首页用纸THESIS：UraniumresourcepriceforecastingbasedonchaostheoryandsupportvectorregressionSPECIALIZATION:ComputationalMathematicsPOSTGRADUATE:GuanLiMENTOR:YanQishengABSTRACTAsthematerialbasisforthedevelopmentofnucl
2、earpower,thechangingruleoftheuraniumresourcesisinfluencedbyfactorssuchaseconomic,political.Studythetendencyoftheinternationaluraniumresourcepricesisimportanttomakingplansondevelopmentandutilizationofnuclearenergy.Atpresent,thepricepredictionresearchisma
3、inlybasedontimeseries,theresearchingmethodsmainlydividedintotwocategories.Oneistraditionalmethod,suchasARMAmodel,GARCHmodelandsoon;anotherisartificialintelligencemethod,suchasneuralnetworkmodel,supportvectormodel,fuzzyregressionmodelandsoon.Althoughthen
4、euralnetworkhasthecharacterofnon-linear,intheapplicationitaffectedbytheinitialvalue,slowingtrainingspeedandeasytofallintolocalminima,andsometimesitoccursover-learningorgeneralizationabilityphenomenon,itwillaffectsthestabilityofthemodelandtheaccuracyoffo
5、recasting.Followedthestructuralriskminimizationprince,theSVMcansolvesomeproblemsbetter,likethesmallsample,nonlinear,high-dimension,localminimumandsoon.Itsolvedthedimensiondisaster,anditisanewresearchinghot-point.Thepapermainlyincludesthefollowingaspects
6、:(1)IntroducedtheChaostheoryandtheMarkovdecisionthought,calculatingtheLyapunovexponenttoexplainthechaoticofthetimeseries,usingtheCaomethodtodecidethesmallestembeddingdimension,andreconstructthetimeseriesofuraniumresources(2)Thisarticledescribesthealgori
7、thm,structureandmodelofsupportvectormachine,introducethetypeofthekernelfunction,whichtheselectionofparametersaffectperformanceandforecastresultsofsupportvectormachinemodel.Inthispaper,usingthetheoryofparticleswarmalgorithmtooptimizetheparametersofsuppor
8、tvectormachinemodel.III万方数据东华理工大学硕士学位论文(3)Onthebasisofthereconstructionoftheuraniumresourcespricetimeseries,builttheuraniumresourcespricepredictionmodelbasedonRBFneuralnetwork,SVRandPSO-SVR,andthencomparingthesimulationexperiment
