城市轨道交通超短时客流预测模型研究及应用

Research and Application of Ultra-short-term Passenger Flow Prediction Model for Urban Rail Transit

超短时客流预测是城市轨道交通调度指挥中的关键基础性问题,现有的方法及模型各有优缺点,尚不能很好地满足现场实际工作需要。首先,基于上海城市轨道交通海量客流数据,对客流特征及其影响因素进行提取与分析,在此基础上引入“K最近邻算法”研究建立超短时客流预测模型。以上海城市轨道交通网络为实际背景的初步应用及结果分析表明,研究成果能对运营当天早晚高峰时段(7:00—10:00和17:00—20:00)客流做出超短时预测,具有较好的准确性、时效性和实用性,为调度指挥提供有力的客流数据支撑,助力构建城市轨道交通网络智慧客运组织调度系统。

Ultra-short-term passenger flow prediction is a key fundamental issue in the dispatch and command of urban rail transit.The existing methods and models have their own advantages and disadvantages,and cannot fully meet the needs of practical work on site.Firstly,based on the massive passenger flow data of Shanghai urban rail transit,this paper extracts and analyzes passenger flow characteristics and influence factors,and introduces the K-nearest neighbor algorithm to study and establish an ultra-short-term passenger flow prediction model.The initial application and result analysis based on the Shanghai urban rail transit network as the actual background demonstrate that the research results have good accuracy,timeliness,and practicality,providing strong passenger flow data(during 7:00—10:00 and 17:00—20:00)support for dispatch and command,and helping to build an intelligent transportation organization and scheduling system for urban rail transit network.