基于多重感应信号的微生物网络协同演化模型与优化方法研究

It is very valuable to construct high efficient swarm intelligent models and optimization methods by utilizing the co-evolution pattern and communication mechanism of the microorganism system, which makes a great significance for developing new artificial intelligence paradigms. In the process, quorum sensing regulation and evolutionary dynamics mechanism are the key issues to design new models and methods. Therefore, the project will conduct a basic research on the behavior dynamics of bacterium network driven by multiple sensing signals and multi-level evolution regulations, aiming to achieve some breakthroughs by overcoming the defects of single evolution mode, fixed swarm properties and insufficient intelligence emergence in existing models. Specifically, we will investigate cooperative response mechanism of bacteria to multiple signals on the basis of designing their ecological attribute parameters. We will introduce multi-level emerging and local self-adaptive theory of complex network into the project to research on typical behavior control strategies and dynamical features of complex bacterium network, including optimal assembling behavior control, dynamic balance of swarm size and network dynamics model with local self-adaptation. Then, we will construct co-evolution models of complex bacterium network and optimization methods, and analyze their efficiency, stability and convergence and other properties. In application, we will validate our models and algorithms by solving practical problems in large scale complex networks. The study will provide a new design mode and important theoretical support on complex system modeling and new swarm intelligent algorithm building.

利用微生物系统的协同进化与通讯机制建立高效的群体智能模型与优化方法对发展新型的人工智能范式有重大意义。而其中的群体感应调控机制与演化动力学机理研究是构建新模型、新方法的关键问题。为此，本项目对多重感应信号驱动的细菌网络行为动力学与多层次演化规律进行基础研究，进而突破传统模型存在的进化模式单一、种群属性固定及智能涌现不足等限制。具体包括：基于细菌生态属性参量模型的设计，研究多重感应信号的协同响应机制；引入复杂网络的多层涌现与局部自适应理论，研究复杂细菌网络的典型行为控制策略及其形成的动力学特征，包括最优聚集行为控制、种群规模动态平衡及局部自适应网络动力学模型；进而，建立复杂细菌网络协同演化模型及优化方法，并对其求解效率、稳定性、收敛性和大规模复杂网络工程的应用性进行性能测试与数学理论分析。本项目的研究成果将对复杂系统建模与新型群体智能算法设计提供理论依据与设计模式。

本项目对复杂网络系统及生物网络的协同演化与群体感应机制进行基础理论与应用研究，提出了基于群体感应的微生物仿生模型，设计了一系列基于菌群仿生行为的进化优化算法，并应用于铜板带生产调度与社交网络影响最大化问题。具体研究内容为：1）提出了细菌群体感应信号协同调控模型，引入基于群体的个体聚集感知行为和生命周期演化模型，包括细菌间沟通机制、个体自适应趋化、自适应调控种群规模机制；2）提出了离散细菌觅食优化算法处理社交网络影响力最大化的三层级联模型；3）为解决大规模高维多目标优化问题，提出基于局部变量分析的多目标优化算法和自适应调整参考向量的多目标优化算法，有效地解决了非规则前沿的多目标优化问题；4）提出了基于学习自动机的自适应头脑风暴优化算法，有效调节算法的探索与开发。通过本项目的研究，已形成了一系列新型的群体智能模型与高效的进化优化方法，发表论文33篇（其中，SCI期刊论文15篇），申请发明专利11项，超额完成了预期研究目标。