基于AFC数据的城市轨道交通短期客流预测实时修正研究

Research on Real-time Correction of Short-term Passenger Flow Forecast of Urban Rail Transit Based on AFC Data

短期客流预测数据一般在预测日前1周或1月内生成,但突发疫情、恶劣天气等非预知情况会引起预测量与实际量有较大偏差,因此需要根据AFC数据对预测趋势及预测指标进行实时修正。首先建立基于AFC数据的客流实时清分推演模型,预测当前时刻前已进站但未出站乘客的出站车站、出站时间,出行路径及所乘车次;然后提出趋势修正策略,根据实时AFC刷卡数据及下一时段的预测OD客流量,采用自下而上的策略对下一时段各车站的预测OD客流量及预测出行记录进行修正;进而采用指标修正策略,将趋势修正与实时清分推演结果结合,生成下一时段全维度修正指标;最后以突发疫情为例,通过实验验证实时修正系统的应用效果。短期客流预测的实时修正能够进一步增强路网运营全貌监视能力及实时调度辅助决策支撑力度。

Short-term passenger flow forecast data is generally generated within one week or one month before the forecast date,but unforeseen situations such as outbreak of epidemic and severe weather will cause a large deviation between the forecast and the actual data.Therefore,it is necessary to correct the forecast trend and forecast indicators in real time according to AFC data.First,a real-time passenger flow clearing and deduction model based on AFC data is established to predict the departure station,departure time,travel route and train number of passengers who have arrived but have not left the station before the current time;then,the trend correction strategy is proposed.According to the real-time AFC card swiping data and the predicted OD passenger flow volume in the next period,the bottom-up strategy is adopted to correct the predicted OD passenger flow volume and predicted travel records of each station in the next period;then,the indicator correction strategy is adopted,and the trend correction is combined with the real-time clearing and deduction results to generate a full-dimensional correction indicator for the next period;finally,taking the outbreak of epidemic as an example,the application effect of the real-time correction system is verified through experiments.The real-time correction of short-term passenger flow prediction can further enhance the overall monitoring capability of railway network operation and the support of real-time dispatching to assist decision-making.