论文部分内容阅读
高维多目标优化问题一般指目标个数为4个或以上时的多目标优化问题. 由于种群中非支配解数量随着目标数量的增加而急剧增多,导致进化算法的进化压力严重降低,求解效率低. 针对该问题,提出一种基于粒子群的高维多目标问题求解方法,在目标空间中引入一系列的参考点,根据参考点筛选出能兼顾多样性和收敛性的非支配解作为粒子的全局最优,以增大选择压力. 同时,提出了基于参考点的外部档案维护策略,以保持最后所得解集的多样性. 在标准测试函数 DTLZ2 上的仿真结果表明, 所提方法在求解高维多目标问题时能够得到收敛性和分布性都较好的解集.
Multi-objective high-dimensional multi-objective optimization problems generally refer to the multi-objective optimization problem when the number of targets is four or more.As the number of non-dominated solutions in the population increases sharply with the increase of the number of targets, the evolutionary pressure of evolutionary algorithms is severely reduced and the solution Low efficiency.Aiming at this problem, this paper proposes a particle swarm high-dimensional multi-objective problem solving method, which introduces a series of reference points in the target space and screened non-dominated solutions that take into account the diversity and convergence based on the reference points Particle global optimization to increase the selection pressure.At the same time, proposed a reference strategy based on the external file maintenance strategy in order to maintain the diversity of the final solution.The simulation results on the standard test function DTLZ2 show that the proposed method in When solving high-dimensional multi-objective problems, the solution set with better convergence and distribution can be obtained.