复杂网络特性对大规模多智能体协同控制的影响

随着分布式多智能体系统应用领域和系统规模的不断扩大,网络特性已成为影响系统性能的一个重要因素.通过研究和分析复杂网络特性对大规模分布式多智能体系统协同控制的影响,对多智能体系统性能的影响做出系统性分析,同时为提出大规模多智能体组织结构的优化算法提供依据.主要针对随机网络、小世界网络、网格网络和无尺度网络这4种典型复杂网络特性,从理论和仿真两方面进行分析.在理论方面,通过基于马尔可夫链的信息传输过程在不同网络结构下的建模,对比分析了信息无偏随机游走模型和智能决策模型下的传输效率.在仿真建模中,主要从智能体间信息传输效率、不同应用领域中集成协同控制效率、对网络故障恢复的影响这3个典型的多智能体系统协同控制应用对比分析复杂网络特性对系统性能的影响.研究结果表明,复杂网络特性如小世界和无尺度特性可以在相同的控制策略下形成明显的性能差异,如果设计合理的控制算法,复杂网络结构将有助于多智能体系统性能的提升.

Effects of Complex Network Characters on the Coordination Control of Large-Scale Multi- Agent System

With the expansion of distributed multi-agent system applications and the increasing scale of the system, the characters of complex network have become an important factor in system performance. This paper makes an initial effort to find the effects of complex network characters on large-scale distributed multi-agent coordination to create a systemic analysis of the system performance and provide organization optimization algorithm designs. The study primarily investigated typical complex networks: random network, small-world network, grid network and scale-free network in multi-agent coordination on theoretical analysis and practical simulations. In theoretical analysis, the study has built the cooperative information transmission model based on Markov chain over different network topologies and compared their efficiencies on either random walk or intelligent routing model. In addition, the study explored the characters of complex network in three main coordination simulations: cooperative information transmission, multi-agent team coordination, and multi-agent network recovery. It is found that the characters of complex network such as small-world or scale-free attributes will bring significant differences in spite of the same coordination schema, and it is feasible to design some desired intelligent algorithms to take the advantage of those effects so that system performance can be promoted.