摘要
为提高多目标优化方法的求解性能,在给出了蚁群算法优化函数类问题求解方法的基础上,提出了基于多蚁群分级优化多目标问题的求解方法。构建了子蚁群以自身启发式信息及以其他子群的启发式信息获得准Pareto解以及采用各子群的每一只蚂蚁获得的准Pareto解作支配判断,从而提高Pareto解的多样性;构建了父蚁群以准Pareto解作为空间节点构成TSP类似的组合优化问题,其求解结果以获得多目标优化问题的Pareto解的前沿,从而提高Pareto解的均匀分布性。通过优化实例验证,结果表明,多蚁群分级优化的多目标求解方法所获得的Pareto解具有解的多样性以及解的均匀分布性。
In order to improve the solving performance of multi-objective optimization problem,this paper proposed a new method based on multi-ant-colony algorithms. Aiming to enhance the diversity of Pareto solutions,quasi-Pareto solutions were constructed by sub-ant-colony algorithm which adopted its own and other sub-ant-colony heuristic information and quasi-Pareto solutions obtained by every ant were used for control judgment. The constructed farther-group ants with the quasi-Pareto solutions which act as space nodes constitute TSP（ traveling salesman problem） ,and then the solutions of the TSP act as the front of solutions for multi-objective optimization problem,hence lead to the enhancement of the uniform distribution of Pareto solutions. Experiment results show that the obtained Pareto solutions by multi-ant-colony optimization based on multi-classification methods have many advantages,such as the diversity and uniform distribution of solutions.
出处
《计算机应用研究》
CSCD
北大核心
2010年第10期3705-3707,3717,共4页
Application Research of Computers
基金
河南省自然科学基金资助项目(2010040818)
河南省教育厅青年骨干教师计划资助项目
河南省教育厅自然基础计划资助项目(2010A520034)
关键词
多蚁群算法
多目标优化
函数优化
动态距离调整
multi-ant-colony algorithm
multi-objective optimization
function optimization
dynamic distance adjusting
