Abstract: In complex rail transit networks, there are always passengers whose trips are relatively fixed. These passengers can act as heuristic "witnesses" and prove the feasibility of a certain trip for other groups of passengers and help to identify potential close contacts with a guaranteed recall rate. This paper aims to develop a method to trace the close contacts in the urban rail transit system. A heuristic tree search method was used to form the feasible trip chains of target passengers by leveraging a limited number of witnesses. It could be determined whether the target passenger was a close contact by identifying whether there were any overlaps between the target trip chain and the infected trip chain. Taking Beijing urban rail transit as an example, volunteers were recruited to ride on specific lines and the information of infected persons was assumed for the study purpose. The effectiveness of the method was verified by extracting relevant Automatic Fare Collection (AFC) data to identify close contacts. In the experimental scenario, the recall rate reached 100% and the accuracy rate was 92.7% using the proposed method, indicating the feasibility of the method. The proposed method is helpful for the relevant department to take appropriate countermeasures to prevent the spread and transmission of the epidemic.

Key words: urban traffic, close contact tracing, witness model, rail transit, heuristic search

摘要： 在复杂轨道交通网络中，某些乘客的出行行程是确定的。这些乘客可以作为启发式的“证人”，为其他乘客证明某个行程的可能性，从而在保证查全率的前提下，准确地找出潜在的密接乘客。本文的目的是开发一套适用于城市轨道交通的疫情密接人群追溯方法，利用有限的确定行程乘客作为虚拟的目击者，采用启发式的树搜索生成目标乘客可能的出行链，通过验证目标出行链与感染者行程是否有交集，判断目标乘客是否为密接乘客。以北京城市轨道交通为例，招募志愿者在特定线路乘车，并假定感染者乘车信息，通过提取有关自动售检票(Automatic Fare Collection, AFC)数据识别密接乘客，以验证方法的有效性。在实验场景下，本文提出的方法对密接乘客的识别查全率达到100%，查准率达到92.7%，表明方法具备一定的可行性。识别结果有助于有关部门针对性采取措施，更高效率地防范疫情蔓延和传播。

关键词: 城市交通, 密接人群追溯, 目击者模型, 轨道交通, 启发式搜索