基于BP神经网络的公共自行车单站点调度需求量研究

发布时间：2018-04-04 03:58

本文选题：城市公共自行车　切入点：车辆调度　出处：《交通运输研究》2016年03期

【摘要】：城市公共自行车租赁系统的合理调度对提高公共自行车使用率至关重要,其关键在于对未来自行车使用情况进行合理预测,确定每个站点调度需求量。通过对站点历史借还车数据与运行特性进行归纳分析,利用BP神经网络模型对单站点借(还)车频次随时间分布的规律进行预测,预测值和真实值平均相差约3辆车,曲线拟合良好,证明模型可实践性较高。在此基础上,按照调度时间窗内站点饱和度动态平衡的原则确定单站点最佳调度需求量。对浙江温州鹿城区街心公园站点的实例分析表明,实行按需调度能使早晚高峰单站点"无位可还"的时间缩短约0.5h以上,从而有效提升站点服务质量和满意度。
[Abstract]:The reasonable scheduling of urban public bicycle rental system is very important to improve the utilization rate of public bicycle. The key lies in making a reasonable prediction of the future bicycle usage and determining the dispatching demand of each station.Through the induction and analysis of the data and operation characteristics of the historical borrowing and returning vehicle at the station, BP neural network model is used to predict the distribution of the frequency of the single station borrowing (returning) vehicle with time. The average difference between the predicted value and the real value is about 3 vehicles.The curve fitting is good, which proves that the model is practical.On this basis, according to the principle of dynamic balance of station saturation in scheduling time window, the optimal scheduling demand of single site is determined.The case study of Jiexin Park in Lucheng District, Wenzhou, Zhejiang Province shows that the implementation of on-demand scheduling can shorten the time of "having no place to return" in the morning and evening peak single station for more than 0.5 hours, thus effectively improving the service quality and satisfaction of the station.
【作者单位】：东南大学经济管理学院;东南大学信息科学与工程学院;东南大学能源与环境学院;
【基金】：国家大学生创新创业训练计划(201510286074)
【分类号】：TP183;U491.225

【相似文献】