第3卷第4期 智能系统学报 Vol 3 Ng 4 2008年8月 CAA I Transactions on Intelligent Systems Aug 2008 随机扰动下多源群体觅食系统建模与仿真 刘佰龙，张汝波，史长亭 (哈尔滨工程大学计算机科学与技术学院，黑龙江哈尔滨150001) 摘要：群集智能是指复杂的集体智能来自简单个体之间以及个体同环境之间的相互作用.通常对群集智能的研究 主要借助于群居生物行为的观察.蚁群觅食行为是研究简单个体产生复杂行为的一个典型的例子，首先建立群体觅 食宏观序参数模型.模型考虑了食物源的量和分布以及环境噪声对个体决策的随机影响.给出2个食物源下系统模 型的数值解，表明在较大的噪声影响下，系统有一定的概率会脱离最优解，到达次优解.在Starbg仿真平台下的实 验结果表明，觅食蚂蚁的数量同任务完成时间以及碰撞频率之间呈现出幂指数关系.这对自组织系统和群集智能的 研究有一定的理论意义，并可以用来指导设计更加有效适应、可靠的智能系统。 关键词：群集智能；自组织行为；觅食模型；Sabp仿真 中图分类号：IP18文献标识码：A文章编号：1673-4785(2008)04034207 Modelng and simulatng the foraging system n multi-source groups with random disturbances L IU Bai-long,ZHANG Ru-bo,SH I Chang-ting (College of Computer Science and Technolgy,Harbin Engineering University,Harbin 150001,China) Abstract:Swam intelligence (SD)is artificial intelligence based on observable collective behavior of decentralized, selforganized systems,otherwise known as social anmals,and is gaining more attention from researchers SI sys- tems are typ ically made up of a population of simp le agents interacting bcally with one another and with their envi- rorment The agents follow very smple rules,and although there is no centralized control structure dictating how individual agents should behave,lcal interactions beteen such agents lead o the emergence of complex glbal be- havior Specially,the foraging behavior of ant cobnies have be viewed as a prototyp ical example of how complex group behavior can arise from smple individual behavors In order study the feature of selforganization in SL, first a macroscop ic serial parametric model for flock foraging was established,in which the richness and distribution of food sources were considered as well as the stochastic effects of a noisy environment Numerical solutions were given for systematic modelswith wo food sources These showed that,in an envirormentwith great noise,the opti- mal solution may not be found and a second-best solution may instead be reached Smulations on the Starlogo plat fom showed a power law relationship beteen the number of ants and completion tme as well as the flux of forag- ers The work presented here may mprove the understanding of selforganization and swamm intelligence It can al- so be used to design more efficient,adaptive,and reliable intelligent systems Keywords:wwam intelligence;selforganization;foraging modeling Starbgo smulation 受群居生物启发的群集智能受到越来越多的研 全局有序的现象被称为自组织.理论上来说自组织 究人员关注.在群居生物中，简单个体之间的相互作来自正反馈和负反馈的平衡.这种非线性现象可以 用涌现出复杂的集体行为.这种低级个体交互产生 借助于数学模型来进行预测.为了解释群集智能中 收稿日期：20080507. 的自组织，对群居生物集体行为的观察是很有必要 通信作者：刘佰龙.Emai止b624@163.cam 的.其中蚁群觅食行为通常被看作研究简单个体产 1994-2008 China Academic Journal Electronie Publishing House.All rights reserved.http://www.cnki.net第 3卷第 4期 智 能 系 统 学 报 Vol. 3 №. 4 2008年 8月 CAA I Transactions on Intelligent System s Aug. 2008 随机扰动下多源群体觅食系统建模与仿真 刘佰龙 ,张汝波 ,史长亭 (哈尔滨工程大学 计算机科学与技术学院 ,黑龙江 哈尔滨 150001) 摘 要 :群集智能是指复杂的集体智能来自简单个体之间以及个体同环境之间的相互作用. 通常对群集智能的研究 主要借助于群居生物行为的观察. 蚁群觅食行为是研究简单个体产生复杂行为的一个典型的例子. 首先建立群体觅 食宏观序参数模型. 模型考虑了食物源的量和分布以及环境噪声对个体决策的随机影响. 给出 2个食物源下系统模 型的数值解 ,表明在较大的噪声影响下 ,系统有一定的概率会脱离最优解 ,到达次优解. 在 Starlogo仿真平台下的实 验结果表明 ,觅食蚂蚁的数量同任务完成时间以及碰撞频率之间呈现出幂指数关系. 这对自组织系统和群集智能的 研究有一定的理论意义 ,并可以用来指导设计更加有效、适应、可靠的智能系统. 关键词 :群集智能 ;自组织行为 ;觅食模型 ; Starlogo仿真 中图分类号 : TP18 文献标识码 : A 文章编号 : 167324785 (2008) 0420342207 M odeling and simulating the foraging system in multi2source groups with random disturbances L IU Bai2long, ZHANG Ru2bo, SH I Chang2ting (College of Computer Science and Technology, Harbin Engineering University, Harbin 150001, China) Abstract:Swarm intelligence (SI) is artificial intelligence based on observable collective behavior of decentralized, self2organized system s, otherwise known as social animals, and is gaining more attention from researchers. SI sys2 tem s are typ ically made up of a population of simp le agents interacting locally with one another and with their envi2 ronment. The agents follow very simp le rules, and although there is no centralized control structure dictating how individual agents should behave, local interactions between such agents lead to the emergence of comp lex global be2 havior. Specially, the foraging behavior of ant colonies have be viewed as a p rototyp ical examp le of how comp lex group behavior can arise from simp le individual behaviors. In order to study the feature of self2organization in SI, first a macroscop ic serial parametric model for flock foraging was established, in which the richness and distribution of food sources were considered as well as the stochastic effects of a noisy environment. Numerical solutions were given for systematic modelswith two food sources. These showed that, in an environmentwith great noise, the op ti2 mal solution may not be found and a second2best solution may instead be reached. Simulations on the Starlogo p lat2 form showed a power law relationship between the number of ants and comp letion time as well as the flux of forag2 ers. The work p resented here may imp rove the understanding of self2organization and swarm intelligence. It can al2 so be used to design more efficient, adap tive, and reliable intelligent system s. Keywords: swarm intelligence; self2organization; foraging; modeling; Starlogo simulation 收稿日期 : 2008205207. 通信作者 :刘佰龙. E2mail: lbl624@163. com. 受群居生物启发的群集智能受到越来越多的研 究人员关注. 在群居生物中 ,简单个体之间的相互作 用涌现出复杂的集体行为. 这种低级个体交互产生 全局有序的现象被称为自组织. 理论上来说自组织 来自正反馈和负反馈的平衡. 这种非线性现象可以 借助于数学模型来进行预测. 为了解释群集智能中 的自组织 ,对群居生物集体行为的观察是很有必要 的. 其中蚁群觅食行为通常被看作研究简单个体产