【作者】 李明；

【导师】 杨洲；

【作者基本信息】 华南农业大学 ， 农业机械化工程， 2019， 博士

【摘要】 传统公交运营服务主要存在线路规划合理性不足、经营模式单一等问题,主要考虑公交的可达率、重复率、覆盖率等指标,而往往忽视乘客的出行需求。随着城市化的不断扩展,传统公交模式已经难以满足人们对快速、便捷、舒适等多样性自由出行的需求。定制公交作为公交的一种延伸服务,基于移动互联网手段以及乘客出行需求而开通设置,弥补了传统公交服务模式的缺陷,是发展多样化、差异化的公共交通服务的新模式,有利于改善城市交通和环境,目前在国内外引起了广泛关注,与之相关的研究也日益成为热点。然而,新生的基于移动互联网的定制公交模式遇到新的问题,包括定制公交客流分析数据基础与传统公交客流分析数据基础存在较大区别、传统公交模式与定制公交模式并轨运行造成不同车辆调度存在一定难题等。为此,本文提出了定制公交客流分析算法以及多目的地定制公交线路运力调配算法,并以直观方式予以展示,从而达到提升定制公交上座率的目的,从而提高定制公交线路规划的合理性和经营效益。具体内容如下:(1)针对传统公交客流分析算法所存在的问题,提出基于时间序列的模糊聚类算法,以进行公交车票预定的历史数据为基础,按照每星期同一天历史相同时段预测未来相同时段的客流分布情况,以达到更科学地进行合理性的线路规划。根据定制公交历史客流情况的特点,结合基于时间序列的模糊聚类算法以相似性搜索实现模糊时间序列的规律发现,通过时间序列数据变换、预测、分类、聚类、相似性搜索等过程,实现了定制公交客流的推导与应用。(2)通达率、重复率等公交运行指标在传统公交运营模式中起决定性作用,但对于定制公交而言,决定性指标转变成出行需求为主的开通要求,并以座位预定、在线支付以及上车验票为主要运营特征。基于直观方式表达乘客出行需求更有利于运营方的线路规划、开通、运营管理工作,为此本文提出了定制公交的出行需求热力图构建算法和实现步骤,结合电子地图进行直观表达。本文系统性地提出出行需求热力图的构建方法与步骤,直观表达了乘客出行需求的分析结果。(3)由于实际运营过程中传统公交运营模式、定制公交模式并轨运行,因此利用传统运力调度算法在两种模式下进行运力调度已不适用。因此本文根据实际情况提出多目的地定制公交线路运力调配算法,达到在传统运营模式、定制公交运营模式同时提供服务的客运车辆根据位置、座位数、排班时间等因素进行有效调度,从而达到高效实用运力的目的。(4)为验证提出方法的可行性与准确性,基于广州市定制公交项目运行的结果进行了统计分析。本章以历史数据作为训练样本,采用2018年3月作为分析数据,传统公交客流预测算法仅取得了53.89%的准确率,而定制公交模式下基于时间序列的模糊聚类算法模型取得了78.04%的预测准确率。在使用多目的地定制公交线路运力调配算法后由于准时性、车辆适用性、发车准点率等用户体验方面的提升,好评人数相对传统方法而言有所提升,好评率达到了85.49%,相对传统调度方法的81.40%提升了4.09%。实验结果表明,本文所建立的算法模型基本能达到运营、管理等方面的数据要求,能够为定制公交线路规划等提供决策支持。更多还原

【Abstract】 The main problems of traditional bus service are lack of rationality of line planning,single operation mode,mainly considering the coverage rate of bus accessibility rate and so on,but often ignore the passenger travel demand and so on.With the continuous expansion and development of the city,the traditional bus mode is difficult to satisfy people’s fast and convenient.The demand for free travel,such as Czechoslovakia,comfort and so on.As an extension service of public transport,the custom public transport is set up based on the mobile Internet and the passenger travel demand.It has made up the defects of the traditional bus service mode.It is a new model for the development of diversified and differentiated public transport services,which is conducive to improving urban traffic and environment.Widespread concern,and related research has increasingly become a hot topic.However,the new problems encountered by the new mode of customizing public transport include the large difference between the data base of the custom bus passenger flow analysis and the traditional bus flow analysis data base,the traditional bus mode and the custom bus mode run parallel to the different vehicle scheduling problems.Therefore,this paper proposes a custom bus passenger flow analysis algorithm and a multi-purpose route scheduling algorithm,which is displayed in a visual way,so as to improve the custom bus seat rate,so as to improve the rationality and operation efficiency of the custom bus line planning.The specific contents are as follows:(1)In view of the problems existing in the traditional bus traffic analysis algorithm,a fuzzy clustering algorithm based on time series is proposed.Based on the historical data of ticket reservation through APP or We Chat,the passenger flow situation in the same period of time is predicted according to the different time period,so as to achieve the purpose of better line planning;according to the custom bus.The characteristics of reservation and temporary booking passenger flow,combining fuzzy clustering algorithm with similarity search to realize the mining of fuzzy time series,through the process of time series data transformation,prediction,classification,clustering,similarity search and so on,the derivation and application of custom bus passenger flow are realized.(2)The change of bus operation mode makes the function of public transportation,such as the overall rate of decision and the repetition rate,gradually decreases in the planning of the bus line,and becomes the mode of travel demand,seat reservation,online payment and inspection ticket.In order to meet the requirements of bus route planning and vehicle scheduling for a large number of passengers,the requirement of the actual operation of the fixed line is fully considered,and the thermal diagram of the custom bus travel demand needs to be set up and expressed in a visual way.This paper systematically proposes the construction method and steps of the travel demand heat map,and intuitively expresses the analysis result of passenger travel demand.(3)As the traditional bus operation mode and the custom bus mode run side by side in the actual operation process,the traditional operation scheduling algorithm is not applicable in the two modes.Therefore,according to the actual situation,this paper puts forward a multipurpose route scheduling algorithm to achieve the purpose of efficient and practical transportation to achieve efficient and practical transportation by the effective scheduling of passenger vehicles which provide service in the traditional operation mode and the custom bus operation mode according to the position,the number of seats,the time of scheduling and so on.(4)In order to verify the feasibility and accuracy of the proposed method,a statistical analysis was carried out based on the results of the customized bus operation in Guangzhou.This chapter takes historical data as the training sample,using March 2018 as the analysis data,the traditional bus passenger flow forecasting algorithm only achieved 53.89% accuracy,and the fuzzy clustering algorithm model based on time series in the custom bus mode has achieved 78.04% prediction accuracy.After the use of multipurpose line power scheduling algorithm to improve the user experience,such as punctuality,vehicle applicability,departure point rate and so on,the number of praise has been improved relative to the traditional methods,the rate of praise has reached 85.49%,and the 81.40% of the traditional scheduling method has been raised by 4.09%.The experimental results show that the algorithm model established in this paper can basically meet the data requirements of operation,management and so on,and can provide decision support for the custom bus route planning.更多还原