城市轨道交通网络限流与列车调整协同优化建模与仿真

项目名称： 城市轨道交通网络限流与列车调整协同优化建模与仿真

项目编号： No.61473210

项目类型： 面上项目

立项/批准年度： 2015

项目学科： 自动化技术、计算机技术

项目作者： 江志彬

作者单位： 同济大学

项目金额： 79万元

中文摘要： 开放时变特征环境下的城市轨道交通网络限流与列车调整协同优化是一个复杂的NP-hard问题，传统基于运筹学和人工智能的分析方法无法满足复杂网络中高强度客流与列车交互的实时决策要求。本项目以由乘客流动和列车运行组成的动态网络为研究对象，以保障网络运营安全和高效为目标，基于网络规模大、因素多、约束和目标多、强实时性和动态性等特点，通过对网络列车运行与乘客出行在时间与空间上的影响与协调机理解析，建立分层递阶式的分布式多智能体架构，重点探讨具有干扰吸收和自适应能力预防反应式鲁棒Agent调度模型、基于数据包络分析的Agent多目标评估模型以及基于强化学习的智能Agent调度规则的自学习模型，最终形成协同优化的相关理论、方法和原型系统。本研究对揭示城市轨道交通网络不确定环境下乘客出行和列车运行的动态匹配机理、提高大客流条件下的限流与列车调整协同优化处置的水平具有重要的理论和实际应用意义。

中文关键词： 城市轨道交通网络；限流措施；列车运行调整；多智能体；调度指挥

英文摘要： The optimization of interaction between passenger holding and train rescheduling in urban rail transit network under complex and changing environment is a NP-hard problem, and traditional analysis method based on the operational research and artificial intelligence cannot satisfy real-time and decision-making requirements of the complex network with interaction of huge passenger flow and trains. Taking the dynamic network consisting of passenger flow and train operation as research object, the project establishes a multi-level hierarchical and distributed multi-agent framework which aims to ensure safety and efficiency of network operation by analyzing the interaction effect of train operation and passenger travel both in time and space, depending on the characteristics of large network scale, multiple factors, constraints and objectives as well as the instantaneity and dynamics. In the project the prevention reactive robustness-agent scheduling model with disturbance absorption and adaptive ability, multi-objective agent evaluation model based on data envelopment analysis as well as self-learning model based on agent scheduling rule with reinforcement learning are discussed. Eventually the multi-agent modeling and simulation theory, method and prototype system optimization of interaction between passenger holding and train rescheduling under the dynamic network are formed. The research has an important role in revealing the dynamic matching mechanism of passenger travel and train operation under the uncertain environment of urban rail transit network, and improving the collaborative optimization of passenger holding and train adjustment under emergency conditions.

英文关键词： rail transit network;passenger holding;train rescheduling;multi-agent;train dispaching