WBMOAIS: Anovelartificialimmunesystemformultiobjectiveoptimization

This studypresentsanovelweight-basedmultiobjectiveartificialimmunesystem(WBMOAIS)based on opt-aiNET,theartificialimmunesystemalgorithmformulti-modaloptimization.Theproposedalgo- rithm followstheelementarystructureofopt-aiNET,buthasthefollowingdistinctcharacteristics:(1)a randomly weightedsumofmultipleobjectivesisusedasafitnessfunction.Thefitnessassignmenthas a muchlowercomputationalcomplexitythanthatbasedonParetoranking,(2)theindividualsofthe population arechosenfromthememory,whichisasetofelitesolutions,andalocalsearchprocedure is utilizedtofacilitatetheexploitationofthesearchspace,and(3)inadditiontotheclonalsuppression algorithm similartothatusedinopt-aiNET,anewtruncationalgorithmwithsimilarindividuals(TASI) is presentedinordertoeliminatesimilarindividualsinmemoryandobtainawell-distributedspread of non-dominatedsolutions.Theproposedalgorithm,WBMOAIS,iscomparedwiththevectorimmune algorithm (VIS)andtheelitistnon-dominatedsortinggeneticsystem(NSGA-II)thatarerepresentativeof the state-of-the-artinmultiobjectiveoptimizationmetaheuristics.Simulationresultsonsevenstandard problems (ZDT6,SCH2,DEB,KUR,POL,FON,andVNT)showWBMOAISoutperformsVISandNSGA-IIand can becomeavalidalternativetostandardalgorithmsforsolvingmultiobjectiveoptimizationproblems.