基于自适应多种群策略的混合多目标优化算法

Adaptive multipopulation strategy based hybrid multiobjective optimization algorithm

为了能够快速准确地获得多目标优化问题的一组具有较好质量和分布性的非支配解,提出了一种自适应多种群混合多目标优化算法.该算法将多目标优化问题分解为多个单目标子问题,在每次迭代时,根据种群在目标空间和解空间的分布情况为多个子问题分别构造子种群,并采用粒子群优化算法对子问题最优解实施搜索,利用差分进化算法对外部档案实施进化.通过对标准测试函数仿真实验,并与经典的及类似策略的多目标优化算法进行比较,结果表明所提出的算法能够利用较少的估值次数获得较好质量和分布性的非支配解集.

In order to obtain a nondominated solution set with higher quality and better distribution quickly and accurately, an adaptive multipopulation strategy based hybrid multiobjective optimization algorithm is proposed. The proposed algorithm decomposes the multiobjective optimization problem into multiple single objective subproblems, and multiple subpopulations are constructed for each subproblem according to the distribution of population in the objective and solution space at each iteration. The particle swarm optimization is used to search the optimal solutions of subproblems, and the external archive is evolved by the differential evolution algorithm. The simulation of several benchmark test functions shows that, compared with the state-of- art and similar multiobjective optimization algorithms, the proposed algorithm can obtain better nondominated solutions with less fitness evaluation times.