摘要
为了提高交叉口运行效率,提出了一种基于胞映射的交叉口自学习模糊控制策略。该方法首先以交叉口各方向排队长度为状态构成状态空间,并进一步将该空间划分为离散的模糊胞元;然后通过研究交叉口交通状态在状态空间内各胞元间的跳转关系,分析交叉口系统的动态特性;最后以此为基础制定模糊控制规则,设计模糊控制器以确定交叉口的绿信比,并引入自学习策略对控制效果进行评估和持续改进控制器性能。基于北京市地安门的实测交通数据的仿真结果表明:该方法与固定配时控制与感应控制相比较,能够明显地减少交叉口各方向的排队长度。
A learning-based fuzzy control strategy was developed using cell mapping to improve intersections＇ efficiency.Queue lengths of every direction at an intersection are regarded as traffic states to form the state space,which is further divided into discrete fuzzy cells.The transition patterns of traffic states transiting among the fuzzy cells are then used to analyze the intersection dynamic characters.Fuzzy control rules are derived based on the dynamic characters with fuzzy a controller designed to optimize the split at the intersection.Self-learning strategy is introduced to keep on improving the controller＇s performance.Simulations based on the data of the Di＇anmen intersection in Beijing demonstrate that the strategy leads to a shorter queue length than the fixed-time and vehicle actuated methods.
出处
《清华大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2010年第5期709-713,共5页
Journal of Tsinghua University(Science and Technology)
基金
国家"九七三"重点基础研究项目(2006CB705506)
国家"八六三"高技术项目(2007AA11Z215)
国家自然科学基金资助项目(60834001
60774034
60721003
50708055)
北京市科委博士生论文资助专项(ZZ0807)
关键词
交叉口控制
胞映射
模糊控制
自学习
交通状态空间
intersection control
cell mapping
fuzzy control
self-learning
traffic state space
