摘要
为了克服蚁群算法难以直接处理连续优化问题的缺陷,在保持蚁群算法基本框架的基础上,将传统蚁群算法中蚂蚁由解分量的信息素和启发式的乘积值按比例来决定取值概率的方式,改为根据连续的概率分布函数来取值.并将函数在各个维上的极值点方向作为蚂蚁搜索的启发式信息.在标准测试函数上的试验结果显示,该算法不但具有较快的收敛速度,而且能够有效地提高解的精确性,增强了算法的稳定性.
A new approach was proposed for solving continuous optimization problems using ant an colony optimization(ACO) algorithm. The method maintains the framework of the classical ant colony algorithm, and replaces discrete summation by the continuous integral, and replaces discrete frequency distribution by continuous probability distribution in the ant selecting probability formula. The direction towards the maximum in each dimension was used as the heuristic information guiding the ants' searching. Experimental results on benchmarks show that our algorithm not only has faster convergence speed but also effectively improves the accuracy of solution and enhances its robustness.
出处
《山东大学学报(工学版)》
CAS
北大核心
2009年第6期24-30,共7页
Journal of Shandong University(Engineering Science)
基金
国家自然科学基金资助项目(60673060
60773103)
江苏省自然科学基金资助项目(BK2008206)
江苏省教育厅自然科学基金资助项目(08KJB520012)
关键词
蚁群算法
约束优化问题
连续函数
ant colony optimization(ACO)
constrained optimization problems
continuous function