基于动态参照点的多主体有限理性路径选择模型  

Dynamic Reference Points based Bounded Rational Multi-Agent Model of Route Choice

作　　者：李雪岩[1] 李雪梅[1] 李学伟[1] 赵云[1] 邱荷婷 

机构地区：[1]北京交通大学经济管理学院,北京100044

出　　处：《复杂系统与复杂性科学》2016年第2期27-35,共9页Complex Systems and Complexity Science

基　　金：国家自然科学基金(71273023);高等学校博士学科点专项科研基金:(20130009110020);中央高校基本科研业务费专项资金(2013YJS039;2014YJS059)

摘　　要：为了研究有限理性假设下出行者的自适应调整行为对交通网络分流的影响,利用累积前景理论结合演化元胞自动机建立了具有个体交互机制的多主体路径选择模型。在模型中将出行者划分为风险追求者与风险厌恶者,基于出行时间可靠性并借鉴元胞遗传算法的思想设计了具有异质特点的出行者动态参照点及其演化规则,使出行者个体能够依据决策环境的变化动态地调整自身的出行时间预算,更加符合出行者的实际行为特征。最后将多主体参照点演化规则与传统的相继平均算法相结合,求解路网配流。研究发现:演化模型较好地继承了传统模型中的路径分流特点;不同的出行者类型比例及出行者的信息接收程度是影响路网分流结构的重要因素。For the research of the impacts of travellers＇ adaptive behavior on traffic flow assignment under bounded rationality, the multi-agent model of route choice with interaction among travellers is established using cumulative prospect theory and cellular automaton, in which travellers are grouped into two types： risk lovers and risk averse. Travellers＇ heterogeneous dynamic reference points and evolution rules are designed based on travel time reliability and the idea of cellular genetic algorithm, so travellers can dynamically adjust their budget of travel time accord- ing to environment. The new model is more in tune with travellers~ actual behavior. Then by combining multi-passengers~ evolution rule with method of successive average, the new traffic flow assignment is solved. The study found that （1） the new model inherited the characteristics of the traditional traffic flow assignment model~ （2） proportions of travellers with different risk attitude and travellers ~ information receiving degree are critical factors which affecting traffic flow assignment.

关 键 词：有限理性 多主体 风险 元胞遗传算法 动态参照点 交通流量分配 

分 类 号：N945[自然科学总论—系统科学]