摘要
微粒群算法已经成功地应用到优化方面,但并不是对所有的约束函数都有效。本文用五种方法来验证该观点,通过对测试函数结果的分析可以看到哪一种方法是最优的或是最坏的;采用五种惩罚策略函数方法,每一方法用五个微粒群算法来测试,从而得出哪一种方法和微粒群算法结合是最好的,并对这一直观的结论进行了理论上的解释。
Particle swarm optimization has been successtully applied to optimization, but it is not eHecuve on au of the constraint functions. This paper validates it through five measures, and analyzes the results of the test function in which we can find which is the best or the worst one, then we use five methods for the punishment function and each of them is tested by five particle swarm optimization methods, and we can decide which one is the best to be combined with particle swarm optimization, and it is explained theoretically.
出处
《计算机工程与科学》
CSCD
北大核心
2009年第5期81-83,86,共4页
Computer Engineering & Science
关键词
微粒群算法
罚函数
正态分布
偏离度
统计算法
particle swarm optimization
penalty function
normal distribution
deviation
statistical algorithms
