摘要： 为了求解带有约束条件的多目标函数优化问题，提出基于连续空间优化的多目标蚁群遗传算法。针对多目标优化问题的特点，定义连续空间中利用信息量指导遗传搜索策略和信息更新方法，将信息量指导遗传搜索、优秀决策引入、决策集更新、改变算法终止条件等方式相结合，有效地加速了搜索的收敛速度，控制了Pareto最优决策集的数量，扩大了决策的分布范围，维持了决策的多样性。数值实验说明该算法能够快速找到一组分布广泛的Pareto最优决策。

关键词: 约束多目标优化, 蚁群遗传算法, Pareto最优决策

Abstract: In this paper, a new algorithm called the multi-objective ant-genetic algorithm, which is based on the continuous space optimization, is presented to solve constrained multi-objective optimization problems. For the trait of multi-objective optimization, the strategy of inheritance searching under the instruction of pheromone and the method of updating pheromone are defined. Then three means are combined together so that the constringent speed of searching has been improved a lot and the quantity of Pareto optimal decisions are controlled, also the distributing area of decisions is enlarged, the diversity of the swarm is maintained. The termination conditions of multi-objective ant-genetic algorithm are presented. An example is listed to prove that the algorithms are effective, and it can find a group of widely distributed Pareto optimal decisions.

Key words: constrained multi-objective optimization, ant-genetic algorithm, Pareto optimal decisions

中图分类号: 

• TP18