引入群体发现和加入行为的随机搜索算法  

A Novel Adaptive Stochastic Search Algorithm Based on Group Founding and Joining Behaviors

作　　者：王虹[1] 卫军胡[1] 刘昌军[1] 曹建福[1] 

机构地区：[1]西安交通大学机械制造系统工程国家重点实验室,西安710049

出　　处：《西安交通大学学报》2013年第12期43-49,共7页Journal of Xi'an Jiaotong University

摘　　要：针对自由搜索算法以及自适应随机搜索(ASS)算法效率低和寻优能力不足的问题,提出了一种基于群体发现和加入行为的随机搜索(HASS)算法。HASS算法采用搜索半径的自适应调整策略提高搜索效率,利用区域混合搜索策略引导群体中的不同个体分别进行全局和局部搜索,通过状态评估方法引入变异策略避免陷入局部最优。HASS算法与ASS算法的主要区别体现在解的选择机制和搜索策略上,HASS算法同时接受较优个体和较差个体,分别为两类个体设计不同的寻优策略来指明搜索方向,增强了算法跳出局部最优的能力和寻优效率。对12个标准测试函数的实验结果表明,该算法的寻优成功率可达100%,较之其他4种算法具有更快的收敛速度和更强的全局搜索能力,特别适于处理复杂的函数优化问题。An adaptive stochastic search algorithm with hybrid strategy （HASS） is presented to improve the low search efficiency and the incompetitive optimization of the free search algorithm and the adaptive stochastic search algorithm （ASS）. The algorithm is based on group founding and joining behaviors that exist widely in nature. The strategy to adaptively update the search radius of each individual is used to improve the search efficiency, and a hybrid search strategy is designed to guide different particles to respectively conduct global or local searches. A new mutation operator is introduced to the evolutionary state estimation of the involved particles to improve the population diversity and to avoid premature convergence effectively. The main between HASS and ASS are in selection mechanism of solutions and the search ＂he experimental results on twelve classic benchmark functions show that the HASS las competitive performance to other four existing algorithms in terms of accuracy, and convergence speed, especially for high-dimensional multimodal problems. differences strategy. T algorithm h robustness

关 键 词：群集智能 自由搜索算法 发现者加入者模型 区域混合搜索策略 

分 类 号：TP13[自动化与计算机技术—控制理论与控制工程] TP18[自动化与计算机技术—控制科学与工程]