地铁站人员疏散路径选择行为分析及Logit建模

Analyses and Logit Models of Route Choice Behaviors During Metro Station Evacuation

掌握地铁站的密集人群在应急疏散过程中的路径选择行为规律,对于紧急情况下保障地铁站人群安全具有重要意义。但现有针对疏散路径选择行为的研究多局限于主观分析与优化算法设计,尚缺乏结合实测数据的客观机理和建模研究。为此,基于地铁站疏散演习数据,重点针对疏散路径中的3个路径选择决策点,提取了这3个位置每个可选路径方向上的人流量、疏散引导信息、人员行为以及每条路径的属性数据,分析了有引导和无引导情况下的人员疏散路径选择行为规律。发现疏散人员在选择路径时具有跟随性,且更倾向于选择人流量大的路径。在本次演习中,3个有疏散引导的方向上,疏散人员对引导的服从率分别为47.25%,51.35%和84.97%,表明了疏散引导人员的关键作用;基于疏散演习数据,构建了包含6个影响因素的人员疏散路径选择行为的Logit模型,并对每个路径选择决策点分别进行了模型参数逻辑回归求解,结果表明,模型都能够通过卡方检验,整体预测正确率分别为76.1%,86.2%和91.2%。模型可直接计算得到在该疏散演习条件下,3个路径选择决策点处每个疏散人员根据实时情况作出判断后选择每条路径的概率。对于微观人群疏散模拟来说,该方法对预测真实情况下人员路径选择和疏散时间具有明显的价值。

Understanding route choice behaviors during emergency evacuation of metro stations is important for the safety of people in a case of emergency.However,existing studies on route choice behaviors in emergency care more about algorithm designs or subjective analyses,while underline behavioral mechanism and modeling based on actual observations is still lack of exploration.Based on the data of a metro station evacuation drill,the route choice behaviors of evacuees at three of decision positions for route choice are analyzed.Data of crowd evacuation are extracted,including flow of people,guiding information for evacuation,behaviors and characteristics of evacuees for each optional route at the three decision positions.Route choices with and without guiding are considered.It is found that evacuees tend to follow others during route choices and prefer routes with large flows.Guiding information exists in three route directions in this drill,and evacuees′rates of complying of guiding information for these three directions are 47.25%,51.35%,and 84.97%,respectively,which indicates that evacuation guidance playes a key role in the evacuation.A Logit model of route choice behaviors for evacuation is proposed by considering six variables,and Logit models for each of the decision position is developed through Logistic Regression based on the data of the evacuation drill.The Logit models all have passed the Chi square test,and the overall correctness rates are 76.1%,86.2%,and 91.2%,respectively.The models can be used to calculate the probability of each evacuee choosing each route at the three decision positions after evaluating environments under the condition of the evacuation drill.The proposed method is particularly useful for predicting route choices and evacuation time of people under actual emergency situations,especially for microscopic evacuation simulations.