基于伪幂函数的离散粒子群算法及其应用  

Discrete particle swarm optimization algorithm based on pseudo power function and its applications

作　　者：李军军[1] 甘世红[1] 许波桅[2] 

机构地区：[1]上海海洋大学电气工程系,上海201306 [2]上海海事大学基础实验实训中心,上海201306

出　　处：《控制理论与应用》2011年第6期834-838,共5页Control Theory & Applications

基　　金：上海市教育委员会科研创新资助项目(09YZ273);上海市科委"创新行动计划"资助项目(09dz1202400);上海市优秀青年基金资助项目(ssc08022);上海海洋大学博士启动基金资助项目(A–2400-08–0296)

摘　　要：针对概率贪婪离散粒子群算法不能兼顾收敛速度与收敛率的缺点,提出一种基于伪幂函数的离散粒子群算法.该方法对贪婪度函数进行伪幂化处理,提高了较远离散位置的选择概率,降低了较近离散位置的选择概率,能有效避免早熟收敛,提高收敛率.对该算法的性能进行了分析.无线传感网络路由优化结果表明,该算法可以获得较好的优化结果.Discrete particle swarm optimization algorithm based on probability greed method doesn＇t provide good tradeoffbetween the convergence speed and the convergence percentage. To avoid this disadvantage, a new discrete particle swarm optimization algorithm is developed by introducing a pseudo power function to the greed function. The probability of choosing a far discrete position is increased; meanwhile the probability of choosing a near discrete position is decreased. Premature convergence is well avoided and the convergence percentage is enhanced. The properties of the algorithm are analyzed. Tests in route-optimization in wireless sensor networks show better results in the application of the proposed method.

关 键 词：离散粒子群优化算法 伪幂函数 无线传感网络 路由 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]