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.
出处
《控制理论与应用》
EI
CAS
CSCD
北大核心
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)
关键词
免疫网络
多AGENT
动态环境
优化
immune network
multi-agent
dynamic environment
optimization