摘要
根据获取的宁波市民卡消费记录信息、公交线路和公交站点等数据,统计出公交上车客流量信息,以客流量定义阈值,运用费歇最优分割法建立公交线路上高平峰的划分模型;基于上述高平峰的划分结果运用遗传算法建立优化的公交调度模型,并检验模型的稳健性;最后运用BP神经网络建立公交线路上的客流量预测模型。结果表明:在工作日,公交线路上的呈现早高峰和晚高峰状态,周末高峰期集中在下午1点到5点,节假日高峰期出现时间较早、客流量较大且持续时间长;优化的各时段发车间隔结果比实际公交运营发车间隔缩短了5~15 min;公交客流量预测模型预测未来某一天公交线路上的客流量平均误差为14.21%。
Bus passenger flow information is calculated based on the acquired data of Ningbo citizen card consumption records,bus routes,and station information.The threshold is defined by passenger flow.The Fisher optimal segmentation method gives the division method of the peak and peace periods on the bus line.The optimized bus dispatch model is established by the genetic algorithm based on the division results of the peak and peace periods mentioned above,and the robustness of the model is tested.Finally,the BP neural network is used to establish a passenger flow prediction model on the bus line.The results show that the morning and evening peak periods are present on the bus routes on weekdays.The peak period on weekends is concentrated from 1 to 5 PM and the peak period of holidays appears earlier,which has a large passenger flow and lasts for a long time.The optimized departure interval of each period is 5 to 15 minutes shorter than the actual bus operation departure interval.The bus passenger flow forecasting model predicts that the average error of passenger flow on a bus line one day is 14.21%.
作者
王惠敏
杨青萍
林婉婷
朱铭慧
王立洪
WANG Huimin;YANG Qingping;LIN Wanting;ZHU Minghui;WANG Lihong(School of Mathematics and Statistics,Ningbo University,Ningbo 315211,China)
出处
《宁波工程学院学报》
2021年第4期27-33,共7页
Journal of Ningbo University of Technology
基金
国家级大学生创新创业训练计划项目(202011646014)
宁波大学大学生科研创新计划(2020SRIP4008)。
关键词
应用数学
费歇最优分割法
遗传算法
BP神经网络
applied mathematics
Fisher optimal segmentation method
genetic algorithm
BP neural network