城市常规公交车行程时间预测方法  

Travel Time Prediction Method of Urban Conventional Bus

作　　者：杨晓春 韦凌翔[2] 胡化鹏 李盈 胡莹莹 冀俊元 陈慧娴 YANG Xiaochun;WEI Lingxiang;HU Huapeng;LI Ying;HU Yingying;JI Junyuan;CHEN Huixian(Security Department,Yancheng Institute of Technology,Yancheng Jiangsu 224051,China;School of Material Science and Engineering,Yancheng Institute of Technology,Yancheng Jiangsu 224051,China;School of Civil and Transportation Engineering,Nanjing University of Technology,Nanjing Jiangsu 210094,China)

机构地区：[1]盐城工学院安全保卫处,江苏盐城224051 [2]盐城工学院材料科学与工程学院,江苏盐城224051 [3]南京理工大学自动化学院,江苏南京210094

出　　处：《交通节能与环保》2023年第5期96-103,共8页Transport Energy Conservation & Environmental Protection

基　　金：江苏省大学生创新训练计划项目(202110305140P)。

摘　　要：为有效解决城市常规公交车行程时间预测精度不足而导致的公交吸引力下降问题,本文构建了基于PSO优化BP神经网络的城市常规公交车行程时间预测模型。首先,通过文献综述及实测数据分析城市常规公交车辆的运行特征,筛选站点距离、路段所在区域、站点停靠时间、行驶车速、站点间信号灯数量、站点间红灯停留时间作为预测模型的输入变量;其次,在构建基于BP神经网络的城市常规公交车行程时间预测模型的基础上,借助PSO算法的全局搜索能力对BP神经网络预测模型的初始权值和阈值进行优化,并据此设计了基于PSO算法优化BP神经网络的城市常规公交车行程时间预测步骤;最后,以成都市147路常规公交车行程时间为例进行验证。案例研究表明:基于PSO-BP神经网络和基于BP神经网络预测模型均可实现较为理想的预测效果,且基于PSO-BP神经网络预测模型的均方误差和平均绝对百分比误差比BP神经网络预测模型降低了13.2和4.4%,具有更好的预测精度。In order to effectively solve the problem of declining bus attractiveness caused by insufficient prediction accuracy of urban conventional bus travel time,a prediction model of urban conventional bus travel time based on PSO optimized BP neural network was constructed.Firstly,the operation characteristics of urban conventional bus were analyzed through literature review and measured data,and the station distance,road section area,stop time at the station,driving speed,number of signal lights between stations,and red light residence time between stations were selected as input variables of the prediction model.Secondly,the initial weight and threshold of the BP neural network prediction model was optimized with the help of the global search capability of the PSO algorithm,on the basis of constructing the urban conventional bus travel time prediction model based on BP neural network.The prediction steps of urban conventional bus travel time based on the PSO algorithm to optimize the BP neural network were designed accordingly.Finally,the travel time of Chengdu No.147 urban conventional bus was used as an example for verification.The case study shows that both the PSO-BP neural network and the BP-based neural network prediction model can achieve more ideal prediction effects,and the mean squared error and mean absolute percentage error of the PSO-BP neural network prediction model are reduced by 13.2 and 4.4%compared with the BP neural network prediction model,which has better prediction accuracy.

关 键 词：城市公共交通 公交行程时间预测 BP神经网络 粒子群优化算法 

分 类 号：U491.11[交通运输工程—交通运输规划与管理]