考虑轨道交通换乘需求的公交发车时刻优化模型  

Bus Departure Time Optimization with Rail Transportation Transfer Demand Prediction

作　　者：刘倩 孙玉娥 黄河[2] 杜扬[2] LIU Qian;SUN Yu-e;HUANG He;DU Yang(School of Rail Transportation,Soochow University,Suzhou 215131,China;School of Computer Science and Technology,Soochow University,Suzhou 215006,China)

机构地区：[1]苏州大学轨道交通学院,江苏苏州215131 [2]苏州大学计算机科学与技术学院,江苏苏州215006

出　　处：《小型微型计算机系统》2022年第2期430-437,共8页Journal of Chinese Computer Systems

基　　金：国家自然科学基金项目(61873177,U20A20182)资助。

摘　　要：轨道交通与地面公交的接驳换乘优化,是城市公共交通系统提升服务水平的重要建设方向.针对该问题,本文研究了在满足常规乘客乘车需求的前提下,如何通过优化公交发车时刻实现公交与轨道交通的高效换乘.首先,挖掘乘客出行轨迹信息获取轨道交通与常规公交的历史换乘客流量,然后训练多层感知器神经网络对轨道交通站点出站后换乘公交的客流需求进行预测.本文选取一条接驳多个轨道交通站点的常规公交线路作为研究对象,建立考虑换乘客流需求的公交发车时刻优化模型,以常规乘客和换乘乘客等待时间成本以及公交公司运营成本最小为目标,最后设计遗传算法求解模型得到优化后的发车时刻.本文以上海市770路公交为例验证模型的有效性,结果表明优化后的发车时刻与原发车时刻相比,有效减少了沿线轨道交通站点换乘客流的换乘等待时间.The optimization of passenger transfer between rail transportation and bus is an essential part of improving the public transportation system′s service.This paper aims to optimize bus departure time so that efficient transfer between rail transportation and bus can be achieved when meeting regular passengers′needs.Thus,this paper first obtains the historical transfer record by mining the passenger trajectories and then trains a multi-layer perceptron neural network to predict the passenger transfer demand for the bus after rail transportation.For each bus line that contains multiple transfer stations,we use a genetic algorithm to compute an optimized bus departure time,which minimizes the waiting time cost for regular passengers and transfer passengers and the bus company′s operating cost.Taking the bus route 770 in Shanghai as an example,we evaluate the performance of the proposed model.Experimental results show that the optimized bus departure time can effectively reduce the waiting time of transfer passengers at transfer stations compared with the original timetable.

关 键 词：城市公共交通 换乘需求预测 发车时刻优化 遗传算法 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]