模拟生物理想自由分布模型的粒子群算法  

Particle swarm optimization algorithm mimicking biological ideal free distribution model

作　　者：秦全德[1] 李荣钧[1] 

机构地区：[1]华南理工大学工商管理学院,广州510640

出　　处：《控制与决策》2011年第12期1840-1845,共6页Control and Decision

基　　金：国家自然科学基金项目(71071057);高等院校博士学科点专项基金项目(20060561002)

摘　　要：在分析生物觅食行为中资源斑块选择理想自由分布模型的基础上,提出一种新型的粒子群算法——理想自由分布粒子群优化算法(IFDPSO).该算法将所有粒子中3个不重叠的个体最优位置的适应度视为资源斑块的食物质量,根据理想自由分布模型随机分配相应数量的粒子到各资源斑块中.为保证群体的多样性,各资源斑块的群体最优位置保持随迭代次数增加而线性递减的距离.在间隔一定的迭代次数后,将各资源斑块的粒子重新组合.标准测试函数的仿真结果表明了IFDPSO算法的有效性.A novel particle swarm optimization（PSO） algorithm, ideal free distribution（IFD） PSO, is proposed based on the analysis of IFD model, in which, three non-overlapping personal best positions of the particles are selected, and their fitness values are regarded as food quality of resource patch. Particles are randomlY assigned to each resource patch according to ideal free distribution model. Particles in each sub-population search the optima independently in accordance with standard PSO algorithm. In order to guarantee the diversity of the whole population, the best position of each sub-population is set to keep a distance, which linearly decreases with iterations. After a certain number of iterations, all sub-population particles are regrouped. The experimental results of benchmark functions show the effectiveness of IFDPSO algorithm.

关 键 词：粒子群算法 理想自由分布模型 多峰函数 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]