AttheperiodofmygraduatestudymyoverallGPA(3.77/4.0)rankedtop5%inthedepartment.InthesecondsemesterIbecameteacherassistantthatisgiventotalentedandmaturedstudentsonly.ThisyearIwontheAcerScholarshipastheoneandonlycandidateinmydepartmentwhichistheultimateaccoladefordistinguishedstudentsendowedbymyuniversity.PresentlyIamPparingmygraduationthesisandtryingforthehonorofExcellentGraduationThesis.

Researchexperienceandacademicactivity

WhenasophomoreIjoinedtheAssociationofAIEnthusiastandbegantonarrowdownmyinterestformyfutureresearch.In1997IparticipatedinsimulationtooldevelopmentfortheschedulingsysteminProf.Wang’slab.WiththetoolofOpenGLandMatlabIdesignedasimulationprogramfortransportationschedulingsystem.ItisnowwidelyusedbydifferentresearchgroupsinNUST.In1998Iassumedandfulfilledasewageanalysis&disposeprojectforNanjingsewagetreatmentplant.Thiswasmyfirstpracticetoconvertalaboratoryideatoacommercialproduct.

In1999IjoinedthedistinguishedProfessorYu-GengXisresearchgroupaimingatNetworkflowproblemsolvingandHeuristicalgorithmresearch.SoonIwasengagedintheFuDanGeneDatabaseDesign.Mydutywastopickuptheusefulinformationamongdifferentkindsofgenematchingformat.ThroughthecomparisonandanalysisformanyheuristicalgorithmsIintroducedanimprovedevolutionaryalgorithm--Multi-populationGeneticAlgorithm.Bypidingawholepopulationintoseveralsub-populationsthisimprovedalgorithmcaneffectivelyPventGAfromlocalconvergenceandpromotevariousevolutionaryorientations.ItprovedmoreefficientlythanSGAinexperimentstoo.InthesecondsemesterIjoinedtheworkshop-schedulingresearchinShanghaiHeavyDutyTyreplant.Theschedulingwasdesignedfortherubber-makingprocessthatcoverednotonlydiscretebutalsocontinuouscircumstances.TomakeabalancepointbetweenoptimizationqualityandtimecostIproposedaDynamiclayeredSchedulingmethodbasedonhybridPetriNets.Thepracticalapplicationshowedthattheaveragemakespanwasshortenedbyalargescale.Ialsopublicizedtwopapersincorejournalswiththisidea.RecentlyIamdoingresearchintheCompositePredictoftheElectricalPowersystemassistedwiththetechnologyofDataMiningforBaoSteel.ItrytocombinetheDecisionTreewithRecedingOptimizationtoprovideanewsolutionfortheCompositePredictiveProblem.Thisprojectisnowunderconstruction.