基于IC卡信息的公交客流OD推算方法研究
|
摘要
深入了解城市居民的公交出行特征,及时、准确、全面地掌握公交出行数据,有利于公交管理部门做出科学的公交规划和运营决策,保证城市公共交通系统正常高效地运营。随着公交IC卡的普及和应用,在方便了广大乘客的同时也提供了一种新的公交客流调查统计手段。公交IC卡、乘客自动计数系统(APC)等先进技术的应用可以获取相对实时动态的客流信息,同时也为获得乘客出行OD提供重要依据。
论文研究了公交IC卡数据的处理和分析方法,根据一站式IC卡刷卡数据判断了乘客的上下车站点,提出了改进的结构化模型对公交线路客流OD进行估计。综合分析研究了既有的OD推算方法,结合算例用不同方法进行求解,对改进的结构化模型的的参数进行灵敏度分析,并验证了模型的优越性,分析了公交线路的站点吸引特征和客流OD出行特征。
论文首先分析了公交IC卡数据结构,详细研究了乘客出行、公交线路及站点、公交运营调度等公交基本信息。在现有IC卡数据及技术条件下,提出了乘客上下车站点的判断方法。针对上车站点的判断采用了聚类分析法,结合公交运营调度信息进行时间匹配,得到了各站点的上车人数。在此基础上提出了基于单个乘客刷卡数据与基于站点吸引两种下车站点判断方法。同时,综合考虑了乘客的出行距离、站点附近的土地利用性质以及站点吸引特征等因素,基于站点吸引得到了各站点的下车人数。
其次,在已知各站点上车人数的基础上,提出了改进的结构化模型对公交线路客流OD进行估计；结合各站点上下车人数,综合分析研究了概率论模型、Markov链方法、迭代比例拟合(IPF)方法和基于流体动力学理论的Tsygalnitzky方法等既有的OD推算模型和方法。通过具体算例对改进的结构化模型的吸引参数进行灵敏度分析。针对各种模型方法所求结果的对比分析表明,改进的结构化模型在不需要各站点下车人数的情况下可直接得到公交线路客流OD矩阵,具有一定的优越性。特别是对一站式IC卡数据处理更为方便和直接。
最后,论文以单线路公交一日内早高峰时段的IC卡刷卡数据为例,对一站式IC卡刷卡数据进行处理分析,得到了各站点上下车人数,并利用不同方法对站点间的客流OD进行推算。通过对公交线路各站点上下车人数、路段断面流量以及公交线路客流OD分布情况的分析,得到了公交线路的站点吸引特征和乘客出行特征。
Only by understanding city residents'bus trip features deeply and acquiring accurate and comprehensive bus trip data timely, can we make scientific public transport planning and operation decisions and make sure urban public transport system operates efficiently. The universal application of bus IC card in many cities not only brings convenience to passengers, but also provides a new means of passenger flow statistical surveys. The advanced techniques, such as bus IC card and passengers automatic counting systems (APC), can help obtain relative real-time dynamic passenger flow information. Also they can provide important basis for the access to passenger trip OD.
We consider the process and analysis methods of bus IC card data in this paper, and judge passengers'on-off stops based on the one-stop IC card data. We also put forward an improved structural model to estimate the OD matrix of the bus lines passenger flow. At the same time, we analyze the pre-existing methods, and solve a numerical example with different methods, analyze the sensitivity of the improved structural model's attraction parameters, and prove the improved structural model's superiority, analyze the stop attraction features and passengers'trip features of bus lines.
The structure of bus IC card data is analyzed firstly. We also give a detailed analysis of the basic transit information such as passenger trip, bus lines and stops, bus operation and scheduling and so on. With the present IC data and technology conditions, we put forward a judging method of passengers'on-off stop. According to the judging of boarding stops, using clustering analysis, combined with the bus operation scheduling with time matching, we get the boarding passenger number of each stop. Above this, we present methods of judging the alighting stops based on single passenger IC data and stop attraction, respectively. At the same time, considering passengers'trip distance, land use properties nearby and stop attraction characteristic, we get the alighting passenger number of each stop based on stop attraction.
Based on the passenger boarding numbers of each stop, we put forward an improved structural model to estimate the OD matrix of the bus lines passenger flow. Combined the passenger's on-off numbers of each stop, methods such as the structural model, the probability theory model, Markov chain approach, iterative proportional fitting (IPF) method, Tsygalnitzky methods based on the theory of fluid dynamics and so on, are analyzed and compared. We analyze the sensitivity of the improved structural model's attraction parameters. According to different models' result, the improved structural model is superior because it can obtain the OD matrix of bus passenger flow without the information about the alighting passenger number, especial for the case of one-stop IC card data process.
Finally, we take an example of the IC data of a single bus line at early peak periods of a day. After processing and analyzing the one-stop IC data, we get the passengers' on-off numbers of each stop, and obtain the OD matrix between each stop with different methods. At last we get the features of stop attraction and passenger trip characteristic by analyzing the on-off passenger number at each stop, link flow and the bus path passenger distribution.
论文研究了公交IC卡数据的处理和分析方法,根据一站式IC卡刷卡数据判断了乘客的上下车站点,提出了改进的结构化模型对公交线路客流OD进行估计。综合分析研究了既有的OD推算方法,结合算例用不同方法进行求解,对改进的结构化模型的的参数进行灵敏度分析,并验证了模型的优越性,分析了公交线路的站点吸引特征和客流OD出行特征。
论文首先分析了公交IC卡数据结构,详细研究了乘客出行、公交线路及站点、公交运营调度等公交基本信息。在现有IC卡数据及技术条件下,提出了乘客上下车站点的判断方法。针对上车站点的判断采用了聚类分析法,结合公交运营调度信息进行时间匹配,得到了各站点的上车人数。在此基础上提出了基于单个乘客刷卡数据与基于站点吸引两种下车站点判断方法。同时,综合考虑了乘客的出行距离、站点附近的土地利用性质以及站点吸引特征等因素,基于站点吸引得到了各站点的下车人数。
其次,在已知各站点上车人数的基础上,提出了改进的结构化模型对公交线路客流OD进行估计；结合各站点上下车人数,综合分析研究了概率论模型、Markov链方法、迭代比例拟合(IPF)方法和基于流体动力学理论的Tsygalnitzky方法等既有的OD推算模型和方法。通过具体算例对改进的结构化模型的吸引参数进行灵敏度分析。针对各种模型方法所求结果的对比分析表明,改进的结构化模型在不需要各站点下车人数的情况下可直接得到公交线路客流OD矩阵,具有一定的优越性。特别是对一站式IC卡数据处理更为方便和直接。
最后,论文以单线路公交一日内早高峰时段的IC卡刷卡数据为例,对一站式IC卡刷卡数据进行处理分析,得到了各站点上下车人数,并利用不同方法对站点间的客流OD进行推算。通过对公交线路各站点上下车人数、路段断面流量以及公交线路客流OD分布情况的分析,得到了公交线路的站点吸引特征和乘客出行特征。
Only by understanding city residents'bus trip features deeply and acquiring accurate and comprehensive bus trip data timely, can we make scientific public transport planning and operation decisions and make sure urban public transport system operates efficiently. The universal application of bus IC card in many cities not only brings convenience to passengers, but also provides a new means of passenger flow statistical surveys. The advanced techniques, such as bus IC card and passengers automatic counting systems (APC), can help obtain relative real-time dynamic passenger flow information. Also they can provide important basis for the access to passenger trip OD.
We consider the process and analysis methods of bus IC card data in this paper, and judge passengers'on-off stops based on the one-stop IC card data. We also put forward an improved structural model to estimate the OD matrix of the bus lines passenger flow. At the same time, we analyze the pre-existing methods, and solve a numerical example with different methods, analyze the sensitivity of the improved structural model's attraction parameters, and prove the improved structural model's superiority, analyze the stop attraction features and passengers'trip features of bus lines.
The structure of bus IC card data is analyzed firstly. We also give a detailed analysis of the basic transit information such as passenger trip, bus lines and stops, bus operation and scheduling and so on. With the present IC data and technology conditions, we put forward a judging method of passengers'on-off stop. According to the judging of boarding stops, using clustering analysis, combined with the bus operation scheduling with time matching, we get the boarding passenger number of each stop. Above this, we present methods of judging the alighting stops based on single passenger IC data and stop attraction, respectively. At the same time, considering passengers'trip distance, land use properties nearby and stop attraction characteristic, we get the alighting passenger number of each stop based on stop attraction.
Based on the passenger boarding numbers of each stop, we put forward an improved structural model to estimate the OD matrix of the bus lines passenger flow. Combined the passenger's on-off numbers of each stop, methods such as the structural model, the probability theory model, Markov chain approach, iterative proportional fitting (IPF) method, Tsygalnitzky methods based on the theory of fluid dynamics and so on, are analyzed and compared. We analyze the sensitivity of the improved structural model's attraction parameters. According to different models' result, the improved structural model is superior because it can obtain the OD matrix of bus passenger flow without the information about the alighting passenger number, especial for the case of one-stop IC card data process.
Finally, we take an example of the IC data of a single bus line at early peak periods of a day. After processing and analyzing the one-stop IC data, we get the passengers' on-off numbers of each stop, and obtain the OD matrix between each stop with different methods. At last we get the features of stop attraction and passenger trip characteristic by analyzing the on-off passenger number at each stop, link flow and the bus path passenger distribution.
引文
[1]陈鹏.车辆IC卡信息采集系统设计[D].北京：北京科技大学,2003.
[2]于滨.基于IC-卡收费系统的动态交通信息采集研究[D].长春：吉林大学,2003.
[3]戴霄.基于公交IC信息的公交数据分析方法研究[D].南京：东南大学,2006.
[4]陈学武,戴霄,陈茜.公交IC卡信息采集、分析与应用研究[J].土木工程学报,2004,2(2).
[5]罗磊.基于IC卡信息的公交客流空间分布特征分析方法研究[D].南京：东南大学,2009.
[6]周涛,翟长旭,高志刚.基于公交IC卡数据的OD推算技术研究[J].城市交通.2007.5:48-52.
[7]窦慧丽,刘好德,杨晓光.基于站点上下客人数的公交客流OD反推方法研究[J].交通与计算机,2007,2:79-82.
[8]任其亮,彭其渊,谢小凇,李淑庆.公交客流OD矩阵反推法[J].交通运输工程与信息学报,2006,4(3):75-79.
[9]刘翠,陈洪仁.公交线路客流OD矩阵推算方法研究[J].城市交通,2007,5(4):81-84.
[10]夏志浩,王胜奎.用公交车站上下客数推算公交OD分布的方法[J].四川联合大学学报(工程科学版),1997,1(2):42-48.
[11]Chen Xumei,Guo Shuxia,Yu Lei,Bruce Hellinga.Short-term Forecasting of Transit Route OD Matrix with Smart Card Data[C].2011 14th International IEEE Conference on Intelligent Transportation Systerms,2011,1513-1518.
[12]Liu Jianfeng,Li Jinhai, Chen Feng, Zhou Yanqiu.Review on Station-to-station OD Matrix Estimate Model and Algorithm for Urban Rail Transit[C].2010 Second International Conference on Computer Modeling and Simulation,2010,3:149-153.
[13]Zhao Hui,Yu Lei,Guo Jifu,Zhao Nale,Wen Huimin,Zhu Lin.Estimation of Time-Varying OD Demands Incorporation FCD and RTMS Data[J].Journal of transportation systems engineering and information technology,2010,10(1):72-80.
[14]郭婕,公交IC卡通勤乘客OD确定方法研究[D].南京：东南大学,2006.
[15]周晶,张伦珂.利用IC卡数据估计公交OD矩阵的模型及算法[J].系统工程理论与实践,2006.4：130-135.
[16]李文权.公共交通客流OD矩阵推算方法研究[D].南京：东南大学,2010.
[17]Ceder,A.Public Transit Planning and Operation:Theory,modeling and Practice[M].Elsevier, 2007.
[18]Lamond,B,Stewart,N.Bregman's balancing method[J].Transportation Research PartB,1981, 15:239-248.
[19]Simon, J. and Furth, P. G.. Generating a bus route O-D matrix from on-off data[J]. Journal of Transportation Engineering, 111,1985:583-593.
[20]Ben-Akiva, M., Macke, P. P. and Hsu, P. S.. Alternative methods to estimate route-level trip tables and expand on-board surveys[J]. Transportation Research Record, 1037, 1985:1-11.
[21]Ben-Akiva, M.. Methods to combine different data sources and estimate origin-destination matrices[J]. In Transportation and Traffic Theory (N. H. Gartner and N. H. M. Wilson, eds),1987,459-481, Elsevier Ltd.
[22]Nguyen, S., Morello, E. and Pallottino, S.. Discrete time dynamic estimation model for passenger origin/destination matrices on transit networks [J]. Transportation Research,22B, 1988:251-260.
[23]Furth, P. G. and Navick, D. S.. Bus route O-D matrix generation: Relationships between bioproportional and recursive methods[J]. Transportation Research Record, 1338,1992:14-21.
[24]Navick, D. S. and Furth, P. G.. Distance-based model for estimating a bus route origin-destination matrix[J]. Transportation Research Record, 1433,1994:16-23.
[25]Alex Cui.Bus passenger origin-destination matrix estimation using automated data collection systems[D].University of California at Berkeley,1996.
[26]Gur, Y. J. and Ben-Shabat, E.. Estimating bus boarding matrix using boarding counts in individual vehicles[J]. Transportation Research Record, 1607, 1997:81-86.
[27]Wong, S. C. and Tong, C. O.. Estimation of time-dependent origin-destination matrices for transit networks[J]. Transportation Research, 32B, 1998:35-48.
[28]Wong, S. C. and Tong, C. O.. The estimation of origin-destination matrices in transit networks[J]. In Advanced Modeling for Transit Operations and Service Planning(W. H. K. Lam and M. G. H. Bell, eds),2003:287-315. Elsevier Ltd.
[29]Friedrich, M., Mott, P. and Noekel, K.. Keeping passenger surveys up to date:A fuzzy approach[J]. Transportation Research Record, 1735,2000:35-42.
[30]Nuzzolo, A. and Crisalli, U.. Estimation of transit origin/destination matrices from traffic counts using a schedule-based approach [C]. In Proceedings of the AET European Transport Conference Held at Homerton College, PTRC. Cambridge, UK.2001.
[31]Nuzzolo, A., Russo, F. and Crisalli, U.. Transit Network Modelling - The Schedule-Based Dynamic Approach[J]. FrancoAngeli, Italy.2003.
[32]Nuzzolo,A.and Crisalli,U.Estimation of transit origin/estination matrices from traffic counts using a schedule based approach[J].Proceedings of AET 2001,September 2001.Homerton College,Cambridge,9-12.
[33]J.J.Barry,R Newhouser,A Rahbee,S.Sayeda.Origin and Destination Estimation in New York city with Automated Fare System Data[J].Transportation Research Record, 1817,2002:183-187.
[34]Wu Z.X. and W.H.K.Lam.Transit passenger origin-destination estimation in congested transit networks with elastic line frequencies [J]. Annals of Operations Research,2006,144(1):363-378.
[35]Li Yuwei,Michael J.Cassidy.A generalized and efficient algorithm for estimating transit route ODs from passenger counts[J].Transportation Research Part B,2007,41:114-125.
[36]Li Baibing.Markov models for Bayesian analysis about transit route origin-destination matrices [J].Transportation Research Part B,2008:1-10.
[37]Qianqian Zhang,M.S.OD flow estimation for a two-route bus transit network using APC data:empirical application and investigation[D].Graduate School of the Ohio State University,2008.
[38]Martin L.Hazelton.Statistical Inference for Transit System Origin-Destination Matrices[J]. Technometrics,2010,52(2):221-230.
[39]吴祥国.基于公交IC卡和GPS数据的居民公交出行OD矩阵推导与应用[D].济南：山东大学,2011.
[40]赵晖.基于公交IC卡信息的居民出行OD推算研究[D].西安：长安大学,2009.
[41]郭婕,陈学武.公交IC卡乘客上车站点确定方法及其应用[C].第一届中国智能交通年会论文 集,2005.
[42]朱晓宏,丁卫东,孙泰屹.公交客流信息采集的方法与技术[J].城市公共交通,2005,(7).
[43]吴美娥.对公交IC卡数据处理分析应用的探索[D].北京：北京交通大学,2010.
[44]夏火松.数据仓库与数据挖掘技术[M].北京：科学出版社,2004.
[45]苏新宁,杨建林.数据仓库和数据挖掘[M].北京：清华大学出版社,2006.
[46]陈文伟.数据仓库与数据挖掘教程[M].北京：清华大学出版社,2006.
[47]西安市城市综合交通体系规划.中国城市规划设计研究院,2011.3.2.
[48]丁勇,邓天民,肖裕民.一种新的公交乘客上车站点的确定方法[J].重庆交通大学学报,2009,28(1)：121-125.
[49]汤效琴,戴汝源.数据挖掘中聚类分析的技术方法[J].微计算机信息(测控仪表自动化),2003(1).
[50]陈学武,戴霄,杨敏.先进的公交出行数据采集分析方法[C].海峡两岸智能交通运输系统学术研讨会暨第二届同舟交通论坛,2005.
[51]窦庆峰.重庆市主城区公交客流预测研究[D].西安：西安建筑科技大学,2006.
[52]冯树民,李晓东.公交客流OD推算方法研究[C].第八届国际交通新技术应用大会论文集,北京,2004.
[53]朱从坤,丁建霆,陈瑜.公交线路OD反推的结构化模型研究[J].哈尔滨工业大学学报,2005,6:851.853.
[54]王伟,徐吉谦.城市交通规划理论及其应用[M].南京：东南大学出版社,1998.
[55]胡腾波,叶建栲.马尔科夫链模型在GIS数据预测中的应用[J].计算机系统应用,2008.8.
[2]于滨.基于IC-卡收费系统的动态交通信息采集研究[D].长春：吉林大学,2003.
[3]戴霄.基于公交IC信息的公交数据分析方法研究[D].南京：东南大学,2006.
[4]陈学武,戴霄,陈茜.公交IC卡信息采集、分析与应用研究[J].土木工程学报,2004,2(2).
[5]罗磊.基于IC卡信息的公交客流空间分布特征分析方法研究[D].南京：东南大学,2009.
[6]周涛,翟长旭,高志刚.基于公交IC卡数据的OD推算技术研究[J].城市交通.2007.5:48-52.
[7]窦慧丽,刘好德,杨晓光.基于站点上下客人数的公交客流OD反推方法研究[J].交通与计算机,2007,2:79-82.
[8]任其亮,彭其渊,谢小凇,李淑庆.公交客流OD矩阵反推法[J].交通运输工程与信息学报,2006,4(3):75-79.
[9]刘翠,陈洪仁.公交线路客流OD矩阵推算方法研究[J].城市交通,2007,5(4):81-84.
[10]夏志浩,王胜奎.用公交车站上下客数推算公交OD分布的方法[J].四川联合大学学报(工程科学版),1997,1(2):42-48.
[11]Chen Xumei,Guo Shuxia,Yu Lei,Bruce Hellinga.Short-term Forecasting of Transit Route OD Matrix with Smart Card Data[C].2011 14th International IEEE Conference on Intelligent Transportation Systerms,2011,1513-1518.
[12]Liu Jianfeng,Li Jinhai, Chen Feng, Zhou Yanqiu.Review on Station-to-station OD Matrix Estimate Model and Algorithm for Urban Rail Transit[C].2010 Second International Conference on Computer Modeling and Simulation,2010,3:149-153.
[13]Zhao Hui,Yu Lei,Guo Jifu,Zhao Nale,Wen Huimin,Zhu Lin.Estimation of Time-Varying OD Demands Incorporation FCD and RTMS Data[J].Journal of transportation systems engineering and information technology,2010,10(1):72-80.
[14]郭婕,公交IC卡通勤乘客OD确定方法研究[D].南京：东南大学,2006.
[15]周晶,张伦珂.利用IC卡数据估计公交OD矩阵的模型及算法[J].系统工程理论与实践,2006.4：130-135.
[16]李文权.公共交通客流OD矩阵推算方法研究[D].南京：东南大学,2010.
[17]Ceder,A.Public Transit Planning and Operation:Theory,modeling and Practice[M].Elsevier, 2007.
[18]Lamond,B,Stewart,N.Bregman's balancing method[J].Transportation Research PartB,1981, 15:239-248.
[19]Simon, J. and Furth, P. G.. Generating a bus route O-D matrix from on-off data[J]. Journal of Transportation Engineering, 111,1985:583-593.
[20]Ben-Akiva, M., Macke, P. P. and Hsu, P. S.. Alternative methods to estimate route-level trip tables and expand on-board surveys[J]. Transportation Research Record, 1037, 1985:1-11.
[21]Ben-Akiva, M.. Methods to combine different data sources and estimate origin-destination matrices[J]. In Transportation and Traffic Theory (N. H. Gartner and N. H. M. Wilson, eds),1987,459-481, Elsevier Ltd.
[22]Nguyen, S., Morello, E. and Pallottino, S.. Discrete time dynamic estimation model for passenger origin/destination matrices on transit networks [J]. Transportation Research,22B, 1988:251-260.
[23]Furth, P. G. and Navick, D. S.. Bus route O-D matrix generation: Relationships between bioproportional and recursive methods[J]. Transportation Research Record, 1338,1992:14-21.
[24]Navick, D. S. and Furth, P. G.. Distance-based model for estimating a bus route origin-destination matrix[J]. Transportation Research Record, 1433,1994:16-23.
[25]Alex Cui.Bus passenger origin-destination matrix estimation using automated data collection systems[D].University of California at Berkeley,1996.
[26]Gur, Y. J. and Ben-Shabat, E.. Estimating bus boarding matrix using boarding counts in individual vehicles[J]. Transportation Research Record, 1607, 1997:81-86.
[27]Wong, S. C. and Tong, C. O.. Estimation of time-dependent origin-destination matrices for transit networks[J]. Transportation Research, 32B, 1998:35-48.
[28]Wong, S. C. and Tong, C. O.. The estimation of origin-destination matrices in transit networks[J]. In Advanced Modeling for Transit Operations and Service Planning(W. H. K. Lam and M. G. H. Bell, eds),2003:287-315. Elsevier Ltd.
[29]Friedrich, M., Mott, P. and Noekel, K.. Keeping passenger surveys up to date:A fuzzy approach[J]. Transportation Research Record, 1735,2000:35-42.
[30]Nuzzolo, A. and Crisalli, U.. Estimation of transit origin/destination matrices from traffic counts using a schedule-based approach [C]. In Proceedings of the AET European Transport Conference Held at Homerton College, PTRC. Cambridge, UK.2001.
[31]Nuzzolo, A., Russo, F. and Crisalli, U.. Transit Network Modelling - The Schedule-Based Dynamic Approach[J]. FrancoAngeli, Italy.2003.
[32]Nuzzolo,A.and Crisalli,U.Estimation of transit origin/estination matrices from traffic counts using a schedule based approach[J].Proceedings of AET 2001,September 2001.Homerton College,Cambridge,9-12.
[33]J.J.Barry,R Newhouser,A Rahbee,S.Sayeda.Origin and Destination Estimation in New York city with Automated Fare System Data[J].Transportation Research Record, 1817,2002:183-187.
[34]Wu Z.X. and W.H.K.Lam.Transit passenger origin-destination estimation in congested transit networks with elastic line frequencies [J]. Annals of Operations Research,2006,144(1):363-378.
[35]Li Yuwei,Michael J.Cassidy.A generalized and efficient algorithm for estimating transit route ODs from passenger counts[J].Transportation Research Part B,2007,41:114-125.
[36]Li Baibing.Markov models for Bayesian analysis about transit route origin-destination matrices [J].Transportation Research Part B,2008:1-10.
[37]Qianqian Zhang,M.S.OD flow estimation for a two-route bus transit network using APC data:empirical application and investigation[D].Graduate School of the Ohio State University,2008.
[38]Martin L.Hazelton.Statistical Inference for Transit System Origin-Destination Matrices[J]. Technometrics,2010,52(2):221-230.
[39]吴祥国.基于公交IC卡和GPS数据的居民公交出行OD矩阵推导与应用[D].济南：山东大学,2011.
[40]赵晖.基于公交IC卡信息的居民出行OD推算研究[D].西安：长安大学,2009.
[41]郭婕,陈学武.公交IC卡乘客上车站点确定方法及其应用[C].第一届中国智能交通年会论文 集,2005.
[42]朱晓宏,丁卫东,孙泰屹.公交客流信息采集的方法与技术[J].城市公共交通,2005,(7).
[43]吴美娥.对公交IC卡数据处理分析应用的探索[D].北京：北京交通大学,2010.
[44]夏火松.数据仓库与数据挖掘技术[M].北京：科学出版社,2004.
[45]苏新宁,杨建林.数据仓库和数据挖掘[M].北京：清华大学出版社,2006.
[46]陈文伟.数据仓库与数据挖掘教程[M].北京：清华大学出版社,2006.
[47]西安市城市综合交通体系规划.中国城市规划设计研究院,2011.3.2.
[48]丁勇,邓天民,肖裕民.一种新的公交乘客上车站点的确定方法[J].重庆交通大学学报,2009,28(1)：121-125.
[49]汤效琴,戴汝源.数据挖掘中聚类分析的技术方法[J].微计算机信息(测控仪表自动化),2003(1).
[50]陈学武,戴霄,杨敏.先进的公交出行数据采集分析方法[C].海峡两岸智能交通运输系统学术研讨会暨第二届同舟交通论坛,2005.
[51]窦庆峰.重庆市主城区公交客流预测研究[D].西安：西安建筑科技大学,2006.
[52]冯树民,李晓东.公交客流OD推算方法研究[C].第八届国际交通新技术应用大会论文集,北京,2004.
[53]朱从坤,丁建霆,陈瑜.公交线路OD反推的结构化模型研究[J].哈尔滨工业大学学报,2005,6:851.853.
[54]王伟,徐吉谦.城市交通规划理论及其应用[M].南京：东南大学出版社,1998.
[55]胡腾波,叶建栲.马尔科夫链模型在GIS数据预测中的应用[J].计算机系统应用,2008.8.