The Research Of A Multi-objective Evolutionary Algorithm Based On Steady-state

Multi-objective evolutionary algorithms (MOEAs) have become an important tool for solving optimization problems in real-world at present. The latest popular MOEAs are mostly based on Pareto dominance relation-ship, use various Pareto-ranking methods to improve individuals in every generation, finally make them be convergent to Pareto front. This paper proposes a new multi-objective evolutionary algorithm based on steady-state andε—dominance relationship. It uses steady-state method to replace normal environment selection, ensures population diversity byε—dominance relationship, and simplifies initial configuration through auto-adaptiveεstrategy. In the meanwhile, SEMOEA is an elitist algorithm with preserva-tion technique. In the last part of this paper, SEMOEA will be compared with SPEA2, NSGA-Ⅱand IBEA, which are three classic algorithms, via multi-objective 0/1 knapsack problem and ZDT series problems. Although these algorithms all have their own strong points, SEMOEA has the better result and much more various population diversities on the whole.