动态环境的人工免疫网络多Agent优化策略  

Artificial immune network multi-agent optimization strategy for dynamic environment

作　　者：史旭华[1,2] 钱锋[1] 

机构地区：[1]华东理工大学化学工程联合国家重点实验室,上海200237 [2]宁波大学电气自动化研究所,浙江宁波315211

出　　处：《控制理论与应用》2011年第7期921-930,共10页Control Theory & Applications

基　　金：国家杰出青年科学基金资助项目(60625302);国家"973"计划资助项目(2009CB320603);国家科技支撑计划资助项目(2007BAF22B05);国家自然科学基金资助项目(20876044);宁波市自然科学基金资助项目(2011A610173);浙江省自然科学基会资助项目(Y1090548)

摘　　要：基于生物免疫网络的核心思想及多Agent技术,提出了动态环境下的人工免疫网络多Agent优化策略(Dmaopt-aiNet)该策略以搜索动态环境中的全局最优解为目标,引入了邻域克隆选择、邻域竞争和协作操作,并同时对Agent自信度状态作自动调整,在优化策略中采用了双重Agent网络结构、双重变异及动态环境检测策略.理论分析了Dmaopt-aiNet算法具有全局收敛性,实验结果表明该算法对高维动态优化问题具有较突出的优越性,能准确定位动态环境下的最优解,具有较好的搜索效果和效率.Based on the idea of biological immune network and multi-agent technology, an artificial immune network multi-agent optimization strategy for dynamic environment（Dmaopt-aiNet） is proposed. The strategy with the target of global optimization introduces neighborhood clonal selection, neighborhood competition and neighborhood collaborative operators. Simultaneously, self-confidence of each agent can be automatically adjusted. In the optimizing process, some strategies such as double-agent network structure, double-mutation strategy and dynamic environmental monitoring are involved. Theoretical analysis shows that Dmaopt-aiNet algorithm is global convergence. Experimental results and com- parison illustrate that Dmaopt-aiNet in dealing with high-dimensional dynamic optimization problems is more superior and can accurately determines the location of the optimum with good effectiveness and efficiency.

关 键 词：免疫网络 多AGENT 动态环境 优化 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]