交叉口Agent间的多遇协调策略及其参数影响分析

A Multi-interaction Coordination Strategy for Intersection Agents and Analysis of Its Parameter Affection

为信号控制的城市道路交叉口定义一个Agent结构模型,在分析相邻交叉口交通流关联关系的基础上,利用记忆因子δ、学习概率α、交叉口交通流变化概率βi等参数阐述了交叉口Agent间的多遇协调流程。交叉口Agent多遇协调采用部分获利协作策略,其交互策略更多地考虑在获利少于对方时候如何以更加协作的态度进行协调。利用记忆因子δ构建了交叉口Agent多遇历史学习协调算法。以交叉口Agent集合达到协调平衡模式需要的交互次数为性能指标,以数个交叉口相连接的主干道为例分析了δ、α、βi等参数对此策略和算法的协调性能的影响,结果表明交叉口Agent集合达到协调平衡模式需要的交互次数随着α的减少、βi的增加、δ的减少而增加,具有一定的动态环境适应能力和协调能力。

An Agent controller model for signalized intersection was defined.The multi-interaction coordination progress for urban intersection Agents was described based on the traffic flow relationship of adjacent intersections using the parameters such as memory factor δ,learning probability α,and local traffic change probability β_i at each intersection.Part lucrative cooperation strategy which can consider more how to coordinate better when getting less gain than opponent was adopted in the multi-interaction coordination.On the basis of this,multi-interactive history learning coordination algorithm was constructed.How the parameters δ,α and β_i will affect the strategy and the algorithm＇s performance measured by the interaction times of intersection Agents needed to reach a given equilibrium pattern was analysed taking an arterial road composed of a few connected intersections as an example.The results show that the interaction times needed to reach a given equilibrium pattern of coordination increases along with the decrease of α and δ and the increase of β_i.So the proposed strategy and algorithm have some dynamic environment adaptability and coordination ability.