T-CPS环境下基于多Agent免疫协同进化理论的微观交通认知方法研究

项目名称： T-CPS环境下基于多Agent免疫协同进化理论的微观交通认知方法研究

项目编号： No.61304197

项目类型： 青年科学基金项目

立项/批准年度： 2014

项目学科： 自动化技术、计算机技术

项目作者： 李永福

作者单位： 重庆邮电大学

项目金额： 26万元

中文摘要： 信息物理融合系统是国际前沿研究领域，交通信息物理系统(Transportation Cyber Physical Systems,T-CPS)体现了交通物理系统与交通信息系统的深度融合，是下一代智能交通系统发展的重要方向。本项目旨在研究一种适应T-CPS环境的微观交通认知方法。具体而言，本项目通过解析T-CPS环境下典型交通参数的特征，提出基于Bayes估计和混沌理论的多交通参数融合方法；应用多Agent免疫协同进化理论，建立微观交通认知模型，研究模型的稳定性条件和动态特性；设计包含流量和速度的双回路控制结构，研究车辆滑模自适应控制，讨论其Lyapunov稳定性和鲁棒性；最后应用TransModeler软件进行仿真实验，验证本方法的有效性。力争通过本项目的研究，形成一种T-CPS环境下的微观交通认知新方法，掌握道路交通拥堵形成的微观机理，为科学制定缓解交通拥堵的决策提供理论依据。

中文关键词： 交通信息物理系统；微观交通；免疫进化协同；多智能体；

英文摘要： Cyber Physical Systems have drawn considerable attention from the international academic circle and Transportation Cyber Physical Systems(T-CPS) is an important direction of next-generation of Intelligent Transportation Systems(ITS), which reflects the deep fusion of transportation physical systems and transportation cyber systems. The project focuses on the microscopic traffic cognitive method under the T-CPS environment.Specifically, a Bayes estimation and Chaos theory-based multi-parameter fusion is proposed after the typical traffic parameters are extracted; And then the microscopic traffic cognitive model is put forward according to the immune co-evolution multi-agent theory to reveal the microscopic mechanism of congestion; Meanwhile, the stability condition and the dynamic characterists of the model are also discussed; Moreover,a two-loop control structure including vehicle speed and flow of traffic is designed to study the vehicle adaptive sliding mode control. And subsequently, the analysis of the corresponding Lyapunov stability and robustness is conducted; Lastly, the simulation with the TransModeler is carried out to validate the effectiveness of this method. According to the research for this project, an effective microscopic traffic cognitive method under the T-CPS environment will be obtained, a

英文关键词： Transportation Cyber Physical Systems；Microscopic Traffic；Immune Co-Evolution；Multi-agent；