城市轨道交通实时历史数据库分析与设计

发布时间：2018-10-14 16:59

【摘要】：城市轨道交通综合监控系统可增强地铁对各种突发事件的应变能力和提高反应速度,保障旅客生命、财产安全,改进地铁服务质量和服务水平,技术难度高,是国家鼓励自主创新的重点。城市轨道交通综合监控系统的核心技术主要体现在实时数据库、历史数据库等软件技术上,直接影响了综合监控系统的系统规模、功能、性能和可靠性。我国在该技术上的积累薄弱、发展时问短,与国外先进技术水平差距大。因此研究实时数据库、历史数据库等关键技术,形成自主知识产权的城市轨道交通综合监控系统,对于提升轨道交通公共服务能力、提高轨道交通机电系统国产化率、降低建设成本、保障国家交通安全具有重要的理论和现实意义。本文的主要研究内容如下：1、基于我国城市轨道交通综合监控系统的行业需求,完成了城市轨道交通综合监控软件平台的架构设计,包括实时数据库、历史数据库、数据采集服务、大机界面和工程配置5个模块。2、完成了基于内存关系数据库为微内核的实时数据库系统的分析与设计,将内核与面向用户经常改动的业务应用进行隔离,提高实时库的可扩充性,同时内核的稳定也将极大提高实时数据库的稳定性。同时,综合运用相关技术,完成了实时数据库的完全分布式架构、基于优先级裁决的多机冗余、在线组态和进程级故障隔离等几个关键功能需求的分析和设计,这些设计极人的提高了系统的可靠性和易用性。3、完成了可应用于轨道交通综合监控的历史数据库的设计。研究和比较了多种压缩技术,综合采用死区压缩、类旋转门的有损压缩和Snappy无损压缩技术,数据压缩率可达达到36：1,保存全线25个车站1年的历史数据只需1087.5GB,历史数据分散在每个车站保存,每个车站只需43.5GB。解决了目前轨道交通综合监控这种高频采样、海量数据的应用场景下,关系数据库在读写性能和存贮容量上的瓶颈问题。同时针对现代计算机体系架构,研究和设计了高速数据读写。通过高速缓存、数据压缩、跳表、冻结表、磁盘顺序写等技术,实现高速历史数据读写,不带有损压缩数据写入可达15万条／秒,数据读取可达10万条／秒。
[Abstract]:The integrated monitoring and control system of urban rail transit can enhance the adaptability and response speed of subway to all kinds of emergencies, guarantee the safety of passengers' life and property, improve the service quality and service level of subway, and the technical difficulty is high. It is the focus of the country to encourage independent innovation. The core technology of urban rail transit integrated monitoring system is mainly reflected in real-time database, historical database and other software technologies, which directly affect the scale, function, performance and reliability of the integrated monitoring system. The accumulation of this technology in our country is weak, the development is short, and the gap between China and foreign advanced technology is big. Therefore, the study of real-time database, historical database and other key technologies to form a comprehensive urban rail transit monitoring system with independent intellectual property rights will enhance the public service capacity of rail transit and increase the localization rate of the mechanical and electrical system of rail transit. It is of great theoretical and practical significance to reduce the construction cost and ensure the national traffic safety. The main contents of this paper are as follows: 1. Based on the industry demand of China's urban rail transit integrated monitoring system, the framework design of urban rail transit integrated monitoring software platform is completed, including real-time database, historical database, Data acquisition service, mainframe interface and engineering configuration are five modules. 2. The analysis and design of real-time database system based on memory relational database are completed, and the kernel is isolated from the business application which is changed frequently by users. At the same time, the stability of the kernel will greatly improve the stability of the real-time database. At the same time, the complete distributed architecture of real-time database, multi-machine redundancy based on priority decision, on-line configuration and process level fault isolation are analyzed and designed. These designs improve the reliability and ease of use of the system. 3. The design of the historical database which can be applied to the integrated monitoring of rail transit is completed. Several compression techniques are studied and compared, including dead-zone compression, lossy compression of rotary gate and Snappy lossless compression. The compression ratio of data can reach 36: 1, and it takes only 1087.5 GB to save the historical data of 25 stations in the whole line for one year, and the historical data is scattered in each station and only needs 43.5 GB per station. This paper solves the bottleneck problem of reading and writing performance and storage capacity of relational database under the application scenario of high frequency sampling and mass data in the integrated monitoring of rail transit. At the same time, high speed data reading and writing is studied and designed for modern computer architecture. By means of cache, data compression, table hopping, freezing table, disk sequential writing and so on, high speed history data can be read and written. Without lossy compression data can be written up to 150000 per second, and data read can be up to 100000 per second.
【学位授予单位】：中国海洋大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP311.13

【相似文献】