实时公交换乘方案查询系统设计与实现

发布时间：2018-02-27 19:54

本文关键词： 历史数据实时数据公交运行指标 BP神经网络 Neuroph　出处：《北京邮电大学》2015年硕士论文　论文类型：学位论文

【摘要】：随着国家在大中城市越来越倡导绿色出行,并采取了一系列措施来鼓励大家乘坐地铁公交出行,从而缓减城市道路拥堵和减少污染物排放、降低PM2.5,更多的人选择公交作为日常通勤的方式。除了合理布局和优化公交线路外,通过挖掘公交运行数据,给乘客推荐基于实时和历史数据,更加精确的乘车方案也是一种提升公交乘坐质量的方式。论文首先对当前国内外公交换乘方案查询系统的发展状况做了描述,基于实时公交数据的应用大多数都是应用在公交到站时间的预测上,很少有把实时公交数据直接用在公交换乘方案的生成上。于是,论文提出了通过公交运行数据实现实时公交换乘方案查询系统的方案,整个方案首先基于原始公交信息计算得到静态换乘方案,然后对公交运行历史数据和实时数据在预处理之后对神经网络进行训练,通过训练好的神经网络就可以快速地计算静态换乘方案的动态代价值,基于新的代价值,由静态换乘方案生成了动态换乘方案。论文对动态换乘方案生成系统进行了需求分析和实现过程的详细介绍,分析了系统对数据库的需求、需要的功能模块,设计了相关的数据库表、功能模块的程序流程、以及模块调用接口。并对BP多层神经网络做了工程上的融合,将其集成到工程代码中。论文也介绍了动态公交换乘系统某些模块所做的黑盒测试、白盒测试流程和测试结果,验证了动态换乘方案查询在不同查询模式和查询条件下的结果,以及在生成动态换乘方案过程中对公交运行指标的选取策略。测试结果显示,动态方案可以在生成过程中根据查询条件选取实时或者历史公交运行指标,根据不同查询条件生成不同的动态换乘方案。另外,也对系统进行了性能测试,测试了系统的并发性能、方案查询相应时间和系统稳定性。
[Abstract]:As the country increasingly advocates green travel in large and medium-sized cities, it has taken a series of measures to encourage people to travel by subway, thereby reducing urban road congestion and reducing pollutant emissions. Reducing PM2.5, more people choose public transport as the daily commuter. In addition to rational layout and optimization of bus routes, by mining transit operation data, recommend passengers based on real-time and historical data. More precise rides are also a way to improve the quality of public transportation. Firstly, the paper describes the development of the query system of bus transfer scheme at home and abroad. Most of the applications based on real-time bus data are applied to the prediction of bus arrival time. Few of the real-time bus data are directly used in the generation of bus transfer scheme. Therefore, the paper puts forward a scheme to realize the query system of real-time bus transfer scheme through transit operation data. The whole scheme is first based on the calculation of the original bus information to obtain the static transfer scheme, and then train the neural network after the pretreatment of the historical data and real-time data of the bus operation. The dynamic substitution value of the static transfer scheme can be calculated quickly by the trained neural network. Based on the new cost value, the dynamic transfer scheme is generated from the static transfer scheme. In this paper, the requirement analysis and implementation process of the dynamic transfer scheme generation system are introduced in detail, and the requirements of the system to the database, the required functional modules, the related database tables and the program flow of the functional modules are analyzed. And the module call interface. The BP multilayer neural network is integrated into the engineering code. The paper also introduces the black box test, the white box test flow and the test results of some modules of the dynamic bus transfer system, and verifies the results of the dynamic transfer scheme query under different query modes and query conditions. The test results show that the dynamic scheme can select real-time or historical bus operation index according to the query condition during the process of generating dynamic transfer scheme. Different dynamic transfer schemes are generated according to different query conditions. In addition, the performance of the system is tested, the concurrent performance of the system is tested, the query time of the scheme and the stability of the system are tested.
【学位授予单位】：北京邮电大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：U495;U491.17

【共引文献】