一种改进的求解多目标优化问题的进化算法

Improved Evolutionary Algorithm for Multi-objective Optimization Problems

针对多目标优化问题,传统进化算法维护种群多样性的方法主要依赖于共享函数,但其小生境半径难以进行有效地设置。该文提出一种改进的求解多目标优化问题的进化算法,新算法引入了近邻函数准则(NFC),将其用于选择过程,可以从种群中选择出较好的个体,并确保种群的多样性。此外,新算法中融入了一种基于近邻函数准则的Pareto候选集的维护方法,利用这种方法可以有效地维护候选解集中个体的多样性。对所提出的算法,从时间和空间复杂度进行了理论分析。对一组典型优化问题的测试表明:该文提出的算法具有较高的搜索性能,解集分布的多样性与收敛性均较理想。

In multi-objective optimization problems,traditional mechanisms of ensuring diversity in a population rely on sharing function.However,the main problem with sharing is that it requires the specification of a sharing parameter.This paper proposes an improved evolutionary algorithm for multi-objective optimization problems and introduces the neighborhood function criterion（NFC） which is applied to selection process.By using this criterion,good individuals can be distinguished from the population and ensure the diversity of the population.On the other hand,the preservation method for Pareto candidate solution set based on NFC is incorporated into the proposed algorithm.This method can maintain the diversity of Pareto candidate set effectively.The complexity of time and space in the proposed algorithm is analysed.For a set of benchmark problems,the experimental results show that the proposed algorithm can search more effectively and provide good performance both in convergence and in diversity of solutions.