基于出租车承载行为的优化决策调度方法研究
发布时间：2018-09-08 15:38
【摘要】：出租汽车作为城市公共交通重要的出行方式,为居民提供快捷、高效的出行服务,随着城市的快速发展和居民生活水平的提高,出租车行业面临着新的挑战:承包经营模式高额承包费弊端、出租车运营空驶率高、乘客要求更便捷的出行服务等。更加迫切的需要通过有效的出租车调度方法来提高出行服务质量。本文通过对出租车承载行为的分析,提出了基于载客热点的出租车调度方法,来提高城市出租车调度管理系统的科学、有效、合理性,提升市民出行的便捷度及出租车公司的运营效益;降低城市出租车的空驶率及运营成本;减少城市道路交通负荷及因乘客滞留导致的公共安全危害。通过对出租车服务特性、承载行为特性和调度特性的分析,了解其承载行为在时间上的特性指标有出行次数、单次出行耗时和时间空驶率,在空间上的特性指标是出行需求分布、出行距离分布、出行路径偏好。通过统计出租车回传至信息中心的出租车位置、速度、载客状态等数据,进一步分析路网运行状况和出租车服务情况;并对数据采集误差分析,通过SQL数据库剔除冗余数据和错误数据;对电子地图进行拓扑处理,如清除微短线、弧段重叠坐标,删除悬挂弧段,坐标转换等;由于GPS数据和GIS地图数据都存在误差,懫用最短距离算法将出租车行驶轨迹数据匹配到GIS路网数据上。通过对南京市出租车GPS数据的处理,获得南京市出租承载行为时间和空间特征为,出租车GPS数据在空间上具有聚类特征,选用基于划分的K-means聚类算法,并通过Weka数据挖掘平台计算聚类结果,得到不同时间段的出租车上下客聚类中心。介绍了目前常用的出租车调度方法有阶梯调度和响应乘客信息的调度方法,前者过于粗略且不能提供出租车有效行驶路径,后者是完全从乘客角度出发,二者都忽略了空载出租车的需求。本文提出了从出租车司机角度,在空载行驶时向调度中心请求调度的方法,该方法是基于乘客的出行热点,然后对载客热点区域的出租车适应度进行评价,主要评价指标是出租车的饱和率和路网负荷度,在载客热点区域的适应度满足相应条件时建议出租车调往该区域,并通过Dijkstra算法计算出最优巡游路径。
[Abstract]:As an important way of urban public transportation, taxi provides residents with fast and efficient travel services, with the rapid development of the city and the improvement of living standards. The taxi industry is faced with new challenges: the malpractice of high contracting fee in contract management mode, the high empty driving rate of taxi operation, and the passengers' request for more convenient travel service, etc. More urgent need to improve the quality of travel services through effective taxi scheduling methods. Based on the analysis of taxi carrying behavior, this paper puts forward a taxi dispatching method based on hot spot to improve the science, efficiency and rationality of urban taxi dispatching management system. To improve the convenience of public travel and the operating efficiency of taxi companies; to reduce the empty driving rate and operating costs of urban taxis; to reduce the urban road traffic load and public safety hazards caused by passenger detention. By analyzing the characteristics of taxi service, load bearing behavior and scheduling, we can find out that the characteristic indexes of carrying behavior in time include travel times, time consuming and time empty driving rate. The characteristic indexes in space are travel demand distribution, travel distance distribution and travel path preference. Through the statistics of taxi position, speed, passenger status and so on, the running status of the network and taxi service are further analyzed, and the error of data collection is analyzed. The redundant data and error data are eliminated by SQL database, and the electronic map is topologically processed, such as removing microshort lines, overlapping coordinates of arcs, deleting suspended arcs, coordinate conversion, etc., because of errors in GPS data and GIS map data. The shortest distance algorithm is used to match the taxi track data to the GIS road network data. By processing the GPS data of taxis in Nanjing, the time and space characteristics of the bearing behavior of taxi in Nanjing are obtained. The GPS data of taxis have clustering characteristics in space, and the K-means clustering algorithm based on partition is chosen. The clustering results are calculated by Weka data mining platform, and the taxicab cluster center is obtained in different time periods. This paper introduces the common methods of taxi scheduling at present, that is, step dispatching and response to passenger information. The former is too rough to provide an effective taxi route, and the latter is entirely from the passenger's point of view. Both ignore the demand for empty taxis. This paper presents a method of requesting dispatch from the dispatching center from the taxi driver's point of view. The method is based on the passenger's hot spot, and then evaluates the taxi fitness in the hot spot area. The main evaluation index is the saturation rate of taxi and the load degree of the road network. When the fitness of the hot passenger region meets the corresponding conditions, it is suggested that the taxi should be transferred to the region, and the optimal cruise path can be calculated by the Dijkstra algorithm.
【学位授予单位】：重庆交通大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：U492.434
本文编号：2230949
[Abstract]:As an important way of urban public transportation, taxi provides residents with fast and efficient travel services, with the rapid development of the city and the improvement of living standards. The taxi industry is faced with new challenges: the malpractice of high contracting fee in contract management mode, the high empty driving rate of taxi operation, and the passengers' request for more convenient travel service, etc. More urgent need to improve the quality of travel services through effective taxi scheduling methods. Based on the analysis of taxi carrying behavior, this paper puts forward a taxi dispatching method based on hot spot to improve the science, efficiency and rationality of urban taxi dispatching management system. To improve the convenience of public travel and the operating efficiency of taxi companies; to reduce the empty driving rate and operating costs of urban taxis; to reduce the urban road traffic load and public safety hazards caused by passenger detention. By analyzing the characteristics of taxi service, load bearing behavior and scheduling, we can find out that the characteristic indexes of carrying behavior in time include travel times, time consuming and time empty driving rate. The characteristic indexes in space are travel demand distribution, travel distance distribution and travel path preference. Through the statistics of taxi position, speed, passenger status and so on, the running status of the network and taxi service are further analyzed, and the error of data collection is analyzed. The redundant data and error data are eliminated by SQL database, and the electronic map is topologically processed, such as removing microshort lines, overlapping coordinates of arcs, deleting suspended arcs, coordinate conversion, etc., because of errors in GPS data and GIS map data. The shortest distance algorithm is used to match the taxi track data to the GIS road network data. By processing the GPS data of taxis in Nanjing, the time and space characteristics of the bearing behavior of taxi in Nanjing are obtained. The GPS data of taxis have clustering characteristics in space, and the K-means clustering algorithm based on partition is chosen. The clustering results are calculated by Weka data mining platform, and the taxicab cluster center is obtained in different time periods. This paper introduces the common methods of taxi scheduling at present, that is, step dispatching and response to passenger information. The former is too rough to provide an effective taxi route, and the latter is entirely from the passenger's point of view. Both ignore the demand for empty taxis. This paper presents a method of requesting dispatch from the dispatching center from the taxi driver's point of view. The method is based on the passenger's hot spot, and then evaluates the taxi fitness in the hot spot area. The main evaluation index is the saturation rate of taxi and the load degree of the road network. When the fitness of the hot passenger region meets the corresponding conditions, it is suggested that the taxi should be transferred to the region, and the optimal cruise path can be calculated by the Dijkstra algorithm.
【学位授予单位】：重庆交通大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：U492.434
【参考文献】
相关期刊论文 前1条
1 王学慧;陈新;丁立波;杨圣芳;;出租车近距离自主呼叫系统设计[J];交通与计算机;2008年03期
相关博士学位论文 前1条
1 邓中伟;面向交通服务的多源移动轨迹数据挖掘与多尺度居民活动的知识发现[D];华东师范大学;2012年
相关硕士学位论文 前2条
1 贾婷;基于乘客需求及分布的出租车调度方法技术研究[D];成都理工大学;2013年
2 温雅静;基于热点载客区域的出租车应急调度方案研究[D];北京交通大学;2014年
，本文编号：2230949
本文链接：https://www.wllwen.com/kejilunwen/daoluqiaoliang/2230949.html
