Engineering-Research-Activity2016｜SHINSHU UNIV.

29/80

HernanLabElectrical& ElectronicEngineeringIn the FutureAfter GraduationEvolutionarycomputationisusedtofindoptimalsolutionstomulti-objective,largescale,complexoptimizationproblems.Evolutionarycomputationsimulatesevolutionanditisappliedtosolveavarietyofdesigninnovationandsustainabilityproblems.Smartpowergrids,intelligentmobilityandtransportationsystems,andintelligentwatergridsarethreekeysystemstowhichevolutionarycomputationcanbeusedtooptimizetheirdesignandimprovetheirsustainabilitytomeettheneedsofthefuture.Additionally,designinnovationinkeyindustries,suchasautomobileandspaceexploration,areareasinwhichevolutionarycomputationwillplayanimportantrole.IntheLaboratorywelearnaboutreal-worldproblem-solvingandoptimizationbycomputationalmeans.Thisrequiresaclearunderstandingoftheproblem,itsmodeling,andthecreationofaprogramtosolveit.Thereisalsoopportunitiestocollaborateinjointresearchwithindustry.Theseactivitiescomplementstheeducationreceivedintheundergraduatecurseandbroadenstheskillsofengineers.AguirreHernanAssociate ProfessorEcuadorian Engineer. PhD from ShinshuUniversity in 2003. Collaborates actively with in-dustry, promotes international exchange, and conducts joint research with national and international institutions. Evolutionarycomputationusescomputationalmodelsofnaturalevolutiontosolvecomplexproblemsinscienceandengineering.Thecreationofsustainablesystemstomeetourneedswithoutcompromisingfuturegenerationsisasocialchallenge.Itdemandsthedevelopmentofnewtechnologiesandtheredesignofourinfrastructure,balancingenvironmental,economic,andsocietalneeds.Sustainabilityproblemsareintrinsicallycomplex,dynamic,large-scale,spanseveraldisciplines,andrequiremultidisciplinaryeffortsandmethodstosolvethem.Weusethepowerofevolutionarycomputationfordesigninnovationandoptimizationofsolutionstocomplexsustainabilityproblems.Evolutionary Computation for Design Innovation and SustainabilityAguirre, Akimoto, Tanaka LabOptimization,akamathematicalprogramming,istheprocessoffindingtheoptimalsolutiontoaproblem.Whenwefaceanoptimizationintherealworld,theobjectivefunctionisoftenblack-boxandwecannotobtaintheoptimummathematically.Instead,wesearchfortheoptimumbyusinganiterativemethodonacomputer.Inhislab,themaininterestisinstochasticsearchalgorithmsincludingevolutionaryalgorithms.Theyanalyzeoptimizationalgorithmsforbetterunderstanding,designmoreefficientalgorithms,andapplythemtooptimizationintherealworld.Wefaceoptimizationeverywhereinscienceandengineeringsuchasvehicledesign,spacecraftrouting,controllerdesign,proteinsinteractionprediction,etc.Inhislabtheyarecontributingtothefrontiersofscienceandengineeringbydevelopingaunifiedframeworkforblack-boxoptimizationproblems.Theyarealsoapplyingtheirmethodologytorealworldoptimizationssuchaslenssystemdesign,shiproutingmakingfulluseofweatherforecast,etc.Furthercontributionsareofcourseexpected.YouheiAkimoto, Ph.D.He received his Ph.D. from Tokyo Institute of Technology. During his Ph.D., he worked as a JSPS research fellow, then worked for INRIA-Saclayin France as a postdoctoral fellow. Since April in 2013, he works at ShinshuUniversity as an assistant professor.Design, Analysis and Applications of Optimization AlgorithmsDiscussion with students and visiting researchers. Prof. Akimoto works in a group with Prof. Aguirre and Prof. Tanaka. Many international students are studying in the group, and of course welcomed.Thestudentslearnfromtheirresearchinthelabtheprocessofproblemsolving;understandingtheproblem,developingasolutiontotheproblem,programmingtorealizethesolution,evaluatingthesolutionobjectively.Theseskillswillbevaluableaftergraduation.Therearelotsofsituationswheretheknowledgeofoptimizationisusefulsuchascardesignindustryandfinancialindustry.Formulation. Define the problem.•Choose variables x(can be constrained)•Define the objective f(can be noisy, ill-conditioned, non-separable, non-convex, rugged)Optimization. Design or choose a method.while not satisfied: generate candidate solution(s)evaluate f(x) for each solutionupdate internal parametersendreturn current recommendation(s)What exactly do we want to optimize? What variables can we control?Process of optimization.How can we generate better points step by step? What information can we exploit from the problem?Electrical& ElectronicEngineeringIn the FutureAfter Graduation27

※このページを正しく表示するにはFlashPlayer10.2以上が必要です