城轨新线客流成长期进出站量短时预测研究

为准确掌握城轨新线开通初期客流演化态势、提高运输组织合理性,针对新线客流变化不稳定、缺乏历史客流数据等问题,提出城轨新线客流成长期进出站量短时预测方法。通过对新线站点进出站量变化规律的分析,基于改进模糊C均值聚类算法,对考虑客流趋势相似性的城轨站点类型划分方法进行研究,并提出城轨新线站点历史数据库构建方法;基于趋势距离对近邻匹配机制进行优化,并根据多元统计回归对K近邻算法进行改进,提出新线站点客流成长期进出站量短时预测方法;结合广州地铁客流数据,对预测方法的有效性进行验证。研究结果显示:新线站点客流成长期内短时进、出站量平均预测效率较既有方法增加了35.68%、32.23%,预测精度较既有方法增加了38.32%、25.80%。

Short-time Forecast of Entrance and Exit Passenger Flow for New Line of Urban Rail Transit During Growth Period

In order to accurately grasp the evolution of the passenger flow in the initial stage of the new urban rail station,improve the rationality of the transportation organization,in response to the unstable changes of passenger flow and the lack of historical passenger flow data,this paper proposed a short-time forecast method for entrance and exit passenger flow in urban rail new station during the growth period.First,the similarity of passenger flow trend in the new station was explored through analyzing the change of passenger flow.Based on the improved Fuzzy C-Means algorithm,a station type division method was proposed for developing the database construction method,which was used to establish the historical passenger flow database for the new station.To forecast the short-term passenger flow in the new station during the growth period,an improved K-nearest neighbor algorithm was proposed through optimizing the neighbor matching mechanism and using the multivariate statistical regression.The historical passenger flow data of Guangzhou Metro was used to verify the forecasting accuracy and calculation efficiency of the proposed method.The results show that the forecast efficiency of entrance and exit passenger flow is increased 35.68%and 32.23%respectively compared with the existing method,and the forecast accuracy is 38.32%and 25.80%respectively increased than the existing method.