Indicator and reference points co-guided evolutionary algorithm for many-objective optimization problems

Highlights：

• In order to produce better offspring, a binary tournament mating selection using two tournament strategies Pareto dominance and perpendicular distance between solution and reference line to measure convergence and diversity respectively, is adopted.

• Our proposed algorithm with well-distributed reference point can achieve competitive performance on the problem with irregular Pareto front, due to indicator ε + I used in IREA.

摘要

•A new algorithm IREA is proposed through creatively combining indicator ε + I with reference points. IREA takes advantage of strength of indicator ε + I and reference points on the convergence and diversity respectively.•In order to produce better offspring, a binary tournament mating selection using two tournament strategies Pareto dominance and perpendicular distance between solution and reference line to measure convergence and diversity respectively, is adopted.•Our proposed algorithm with well-distributed reference point can achieve competitive performance on the problem with irregular Pareto front, due to indicator ε + I used in IREA.

论文关键词：Many-objective Optimization,Convergence,Diversity,Indicator,Reference points,MaOPs,Many-objective optimization problems,MOP,Multi-objective optimization problem,MOEAs,Multi-objective evolutionary algorithms,NSGAII,Non-dominanced sorting genetic algorithm II,SPEA2,Strength Pareto evolutionary algorithm 2,SPSAII,Pareto envelope-based selection algorithm II,PF,Pareto Front,MOEA/D,Multi-objective evolutionary algorithm based on decomposition,MOEA/DD,MOEA based on dominance and decomposition,IBEA,Indicator-based evolutionary algorithm,HypE,Hypervolume estimation algorithm,PCSEA,Pareto corner search evolutionary algorithm,MVU,Maximum variance unfolding,PCA,Principal component analysis,SDE,Shift-based density estimation,crEA,Clustering-ranking evolutionary algorithm,θ-DEA,θ dominance based evolutionary algorithm,GD,Generational distance,SP,Spacing,HV,Hypervolume

论文评审过程：Received 28 December 2016, Revised 14 August 2017, Accepted 20 October 2017, Available online 23 October 2017, Version of Record 6 December 2017.