轨道交通网络换乘路径选择方法研究

发布时间：2018-06-24 22:44

本文选题：换乘路径优化 + 客流预测　；参考：《北京交通大学》2015年博士论文

【摘要】：在智能交通系统(ITS)领域,交通网络动态客流分配理论是其关键技术。虽然动态客流分配理论在传统公共交通网络中研究和应用比较广泛,但与轨道交通网络中实时客流密集度相结合,提供动态的换乘路径信息方面存在不少亟待解决的问题。交通诱导系统是轨道交通运营管理发展的方向与趋势,如何有效、实时地引导乘客选择优化的出行路径,合理降低早晚高峰期间轨道路网中核心线路区间的客流运营压力,是当前交通系统工程科学研究热点问题之一。针对上述问题,基于北京轨道交通路网的基础客流数据,本文重点研究了短期客流预测、客流密集度指数体系和客流路径引导模型等关键问题,实现了轨道交通网络化运营中的乘客出行路径选择优化的目的。本文的主要研究内容如下：(1)提出了基于支持向量机的短期组合客流预测算法。首先给出了遗传算法与支持向量机组合的客流预测算法,其中遗传算法能对支持向量机的参数选择进行优化,使得组合算法具有更准确的预测效果。其次提出了小波变换与支持向量机结合的客流预测算法,其中小波分解能无损地将客流信息分解为高频和低频数据,并获取多尺度细化的低频序列,然后支持向量机对一个低频和多个高频序列进行预测,最后对预测的多个序列进行小波重构得到最终的客流预测结果。本文以北京轨道交通网络客流数据为基础,在多种标准评价方式下,实证结果表明本文提出的客流预测算法与多种当前比较流行的客流预测算法相比获得更好的客流预测结果。(2)提出了一种基于轨道交通网络客流密集度的路径选择模型。首先针对轨道交通网络的划分层次,提出了区间、线路和全路网客流密集度指数,实证结果表明不同层次的客流密集度指数均能较好地反映轨道交通网络的客流密集度。其次以客流密集度指数为基础,提出了一种基于轨道交通网络客流密集度的路径选择非集计模型。该模型依据路网客流密集度和路网基础数据实时计算路径广义费用,动态调整路径分配比例并模拟路网客流分布。实证结果表明本文提出的路径选择模型能较好地模拟轨道交通网络中的客流动态分布变化。(3)提出了一种路径引导下的时变比例调整模型。首先以客流密集度指数为基础,给出了一种根据交通诱导系统信息实时选择广义费用最小路径的路径引导模型,以达到动态调整轨道交通网络客流压力的目的。其次通过数学推导证明了模型的可行性,并推导验证了模型的多项基本性质。最后以北京轨道交通客流数据为基础,从“区间--线路--路网”三个层次进行模拟,实证结果表明交通诱导信息能较好地引导乘客选择优化的出行路径,降低轨道交通网络中部分线路和区域的客流密集度,合理降低早晚高峰期间核心线路的客流运营压力,实现换乘优化,达到改善路网客流状况的目的。
[Abstract]:In the field of Intelligent Transportation system (its), the theory of dynamic passenger flow assignment in traffic network is the key technology. Although the theory of dynamic passenger flow assignment is widely studied and applied in the traditional public transport network, there are many problems to be solved urgently in providing dynamic transfer path information by combining with the real-time passenger flow density in rail transit network. Traffic guidance system is the development direction and trend of rail transit operation management. How to effectively and real-time guide passengers to choose the optimized travel path and reasonably reduce the passenger flow operation pressure in the core section of rail network during the morning and evening rush hour. It is one of the hot issues in the research of traffic system engineering. In view of the above problems, based on the basic passenger flow data of Beijing rail transit network, this paper focuses on the key issues such as short-term passenger flow prediction, passenger flow intensity index system and passenger flow path guidance model, etc. The purpose of optimizing the passenger travel path in the network operation of rail transit is realized. The main contents of this paper are as follows: (1) A short-term combined passenger flow prediction algorithm based on support vector machine (SVM) is proposed. In this paper, a passenger flow prediction algorithm based on the combination of genetic algorithm and support vector machine is presented, in which the genetic algorithm can optimize the parameter selection of support vector machine, so that the combination algorithm has more accurate prediction effect. Secondly, a passenger flow prediction algorithm based on wavelet transform and support vector machine is proposed, in which wavelet decomposition can decompose the passenger flow information into high frequency and low frequency data, and obtain the low frequency sequence with multi-scale thinning. Then support vector machine (SVM) is used to predict one low frequency and several high frequency sequences. Finally, wavelet reconstruction of multiple predicted sequences is carried out to get the final result of passenger flow prediction. Based on the passenger flow data of Beijing rail transit network, this paper is based on a variety of standard evaluation methods. The empirical results show that the proposed passenger flow forecasting algorithm is better than many popular passenger flow forecasting algorithms. (2) A route selection model based on the passenger flow density of rail transit network is proposed. First of all, according to the hierarchy of rail transit network, the author puts forward the passenger flow intensity index of interval, route and the whole network. The empirical results show that the different levels of passenger flow intensity index can reflect the passenger flow intensity of rail transit network. Secondly, based on the passenger flow intensity index, a path selection disaggregate model based on the passenger flow density of rail transit network is proposed. Based on the density of the passenger flow and the basic data of the road network, the model calculates the generalized cost of the route in real time, dynamically adjusts the distribution ratio of the route and simulates the distribution of the passenger flow in the road network. The empirical results show that the proposed path selection model can well simulate the dynamic distribution of passenger flow in rail transit networks. (3) A time-varying proportional adjustment model under path guidance is proposed. Based on the passenger flow intensity index, this paper presents a real-time route guidance model for selecting the generalized minimum cost path according to the traffic guidance system information, in order to dynamically adjust the passenger flow pressure of rail transit network. Secondly, the feasibility of the model is proved by mathematical derivation, and many basic properties of the model are proved. Finally, based on the passenger flow data of Beijing rail transit, the simulation is carried out from the three levels of "interval-line-road network". The empirical results show that the traffic guidance information can better guide passengers to choose the optimal travel path. The passenger flow density of some routes and regions in the rail transit network is reduced, the operation pressure of the core lines during the morning and evening peak period is reduced reasonably, the transfer optimization is realized, and the passenger flow condition of the road network is improved.
【学位授予单位】：北京交通大学
【学位级别】：博士
【学位授予年份】：2015
【分类号】：U291.75;U495

【相似文献】