基于APC和GPS数据的青奥会期间南京公共交通的调度与优化  

An APC and GPS Data- based Optimization Model for the Dispatch of Public Transportation for the Nanjing Youth Olympic Games

作　　者：武灵艳 邓子豪[1] 吴俣[1] 王昉健 徐金花[2] 王加兵[1] 刘雨田[1] 刘文军[1] 

机构地区：[1]南京信息工程大学数学与统计学院,南京210044 [2]南京信息工程大学信息与控制学院,南京210044

出　　处：《数学理论与应用》2014年第1期116-124,共9页Mathematical Theory and Applications

基　　金：国家自然科学基金(11301277);国家大学生创新训练计划(201310300030)

摘　　要：为了得到青奥会期间南京市合理有效的公交调度方案,本文针对青奥会场馆、运动员村、旅游点等附近的南京公共交通线路,建立模型与算法.首先,通过APC数据与GPS数据的匹配,对客流数据进行站点匹配预处理,根据已有客流量数据,训练小波神经网络,从而对客流分布情况进行预测,然后基于客流预测结果,采用有序聚类法,实现客流高低峰时段的合理划分.其次,详细分析调度问题的关键所在,以时段总发车次数和乘客等待时间两个因素作为目标函数,将时段最大、最小发车间隔和满载率等作为约束条件,提出基于APC和GPS的公交车辆辅助调度模型,通过遗传算法对模型进行求解,得出不同时段的发车间隔和配车次数,并对模型的性能进行评估.以南京市D7路公交运营线路的实际客流数据为例,采用MATLAB软件进行仿真实验,得出优化结果.结果表明所建模型是合理的,从而为调度时刻表的生成提供了科学的依据.This paper aims to build some reasonable and effective public transport scheduling models and correspond- ing algorithm for the Nanjing Youth Olympic Games, according to the Nanjing public transport links nearby Youth O- lympic stadium, athletes village, tourist spots, etc. To begin with, thanks to the pattern of APC data and GPS data transmitting at the same time at the same bus stop, the stop information according to each APC data can be found by map- matching algorithm. With the given conditions and data, we train the Wavelet Neural Network, then the dy- namic change pattern of passenger - flow can be acquired by analyzing the history data, by which the distribution trends can be predicted. By using the sequential clustering method, we also go further to divide the dispatching peri- ods according to the predicted results of distribution of passenger - flow. Secondly, the assistive bus dispatching mod- el based on APC and GPS data is proposed in which the passengers＇ and agency＇ s costs are regarded as the objective function and the departing time interval and the average capacity rate as the constraint conditions. Besides, by using Genetic Algorithm, we get the optimal departing time intervals and number of buses for each period. Also, this paper evaluates the model based on the above results. In the end, as an example, we use MATLAB to do the simulation ex- periment based on the passenger flow data of D7 in Nanjing city. Test results indicate that the established model is practicable and reasonable, and the present work provides a scientific basis for the generation of a bus schedule.

关 键 词：青奥会 公交调度 客流量 有序聚类法 遗传算法 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]