Abstract: By hybrid guidance, a preference multi-objective particle swarm optimization algorithm (HG-MOPSO) which combines the notion of reference points with reference regions is proposed to obtain the optimal effective set in preference regions. In the process of moving reference points, the algorithm dynamically adjusts reference regions to increase the selection pressure and control preference region whose centre is the reference point. Through the improvement of the option modes of gBest of PSO algorithm by spherical sector dominance (ss-dominance) proposed in this paper, the search for Pareto optimal set of multi-objective optimization problems is implemented. Simulation results show that the proposed algorithm is effective.

Key words: Multi-objective optimization, preference regions, particle swarm, hybrid guidance

摘要： 为了获取偏好区域的Pareto解集，将参考点引导方式和区域引导方式结合在一起，提出了一种基于混合引导的偏好多目标粒子群算法(HG-MOPSO)。该算法将参考点作为参考区域的中心，在移动参考点的过程中，动态调整参考区域的大小，增加了选择压力，控制了偏好的范围。另外，改进算法引入球扇占优的概念，优化了全局最优粒子的选取，实现对多目标优化的非劣解有效搜索。仿真结果验证了该算法是有效的。

关键词: 多目标优化, 偏好区域, 粒子群, 混合引导