Utilising the existing infrastructure in railway transit to tackle overcrowding requires more understanding of how people use spaces at stations. This study investigated passenger behaviour while waiting for a train on the platform using the data of the Wi-Fi location tracking systems. The trajectories of 129,354 devices were observed in two weeks at two MRT Circle Line stations in Singapore, which have the escalator/stair landings in different positions. A data cleaning process was proposed to overcome the drawbacks of Wi-Fi-based position data. A decomposition method was further developed to separate the walking and staying phases based on data processing. The boarding passengers’ on-platform behaviour was analysed from four aspects: the number of staying phases, the location distributions of different kinds of stays, the location distribution of in-between stays by hour and duration, and the distance and walking speed of the first walking phase. Our results suggested that many passengers (44% and 37% of passengers at the two case study stations) had multiple staying phases, meaning that they did not go directly to their final boarding points after coming to the platform but rather made stops or walkarounds before coming to boarding points. The distributions of locations of the last and in-between stays were significantly different and may influenced by the width, length and layout (such as landing locations) of stations. In addition, the walking speeds of passengers observed on the metro platform were slower than those observed on the streets. These findings indicated that some commonly used assumptions in most simulation models are not true according to the empirical observation. The obtained knowledge would deepen the understanding of the passengers’ on-platform behaviour and thus provide implications for designing railway stations and planning station operations.

利用铁路运输中的现有基础设施来解决过度拥挤问题，需要更多地了解人们如何使用车站的空间。本研究使用 Wi-Fi 位置跟踪系统的数据调查了乘客在站台上候车时的行为。在新加坡的两个 MRT Circle Line 车站，两周内观测了 129,354 台设备的轨迹，这两个车站的自动扶梯/楼梯平台位于不同的位置。提出了一种数据清洗过程来克服基于 Wi-Fi 的位置数据的缺点。进一步开发了一种分解方法，以基于数据处理分离行走阶段和停留阶段。从四个方面分析了乘客在平台上的行为：停留阶段的数量、不同类型停留的位置分布、按小时和持续时间划分的中间停留位置分布，以及第一个步行阶段的距离和步行速度。我们的结果表明，许多乘客（两个案例研究站的 44% 和 37% 的乘客）有多个停留阶段，这意味着他们在到达站台后没有直接前往最终登车点，而是在到达登车点之前停下来或绕行。最后一次和中间停留的位置分布存在显著差异，并且可能受到车站宽度、长度和布局 （如着陆位置） 的影响。此外，在地铁站台上观察到的乘客的步行速度比在街道上观察到的慢。这些发现表明，根据实证观察，大多数仿真模型中一些常用的假设并不成立。 所获得的知识将加深对乘客在站台上行为的理解，从而为设计火车站和规划车站运营提供参考。