不完全信息下复杂多自主体网络的辨识与性能极限算法研究

项目名称： 不完全信息下复杂多自主体网络的辨识与性能极限算法研究

项目编号： No.61472122

项目类型： 面上项目

立项/批准年度： 2015

项目学科： 其他

项目作者： 姜晓伟

作者单位： 湖北师范大学

项目金额： 80万元

中文摘要： 基于生物智能网络的群体自组织性、局部自治性、涌现性和异质性等特点，本项目针对不完全信息及多种约束下的复杂多自主体网络的辨识算法和性能极限优化计算问题进行研究。基于观测节点的状态信息（位置、速度和方向等），针对信息的不完全性（部分节点状态可测、有限智能造成的局部感知和通信过程中的数据丢失等），挖掘出反映节点和网络动态行为演化规律的网络拓扑结构和重要参数，建立节点和网络的动态演化模型；揭示出多种约束（节点能力约束和通信约束等）和网络拓扑结构、系统参数之间的定量关系，提出复杂多自主体网络在多种约束下的鲁棒辨识算法；结合多机器人合作系统，提炼出具有普适意义的性能指标（跟踪、一致性等），通过优化计算得到复杂多自主体网络在不完全节点状态信息和多种约束下的跟踪或一致性性能极限值，以及跟踪和一致性协同优化算法的设计方法。本项目将为复杂多自主体网络的辨识算法设计和性能极限优化计算提供理论和实际应用上的指导。

中文关键词： 不完全信息；多自主体网络；辨识算法；性能极限

英文摘要： Based on the essential characteristics of biological intelligent network such as self-organization of group, local autonomy, emergence and heterogeneity, this project investigate identification algorithms and optimization computation of performance limitations for multi-agent networks with incomplete information and various constraints. From the state information of observed nodes including positions, velocity and directions, and in view of the incomplete information which may result from state variables are partially observable, local awareness with limited intelligence and data loss in the communication process, we can dig out the network topology and key parameters which reflect the evolution laws of dynamical behavior of nodes and networks to establish dynamical evolution models. By revealing the quantitative relationships among various constraints, network topology, system parameters, we will propose the robust identification algorithms of complex multi-agent networks under kinds of restrictions. Combining with Multi-robot cooperation system, the performance index which have universal significance will be refined, such as tracking and consensus performance index. Then tracking and consensus performance limitations values of complex multi-agent networks under incomplete informations and various constraints can be obtained by using optimization computation, and design methods of collaborative optimization algorithms for tracking and consensus can also be given. This project will give guidance in theory and pratical applications for identification algorithm design and optimization computation of performance limitations.

英文关键词： Incomplete information;Multi-Agent networks;Identification Algorithm;Performance limitations