【作者】 宋志雄；

【导师】 蒲云；

【作者基本信息】 西南交通大学 ， 交通运输规划与管理， 2013， 博士

【摘要】 智能化与计算机、半导体和微电子技术发展密不可分。智能化已在各个领域得到广泛的使用,如智能手机、智能小区、智能车辆和智能交通等。为满足高速增长的铁路运力需求,我国运营的高速列车里程已经世界第一。铁路经历了多次的大面积提速,铁路线路养护还需要依靠大量的人力投入,尤其是铁路大型养路机械(以下简称大机)已经大量投入使用。但是,大机的管理水平、安全状况和工作效率常常受到诟病,如何提高大机的安全性、可靠性和作业效率是当前铁路工务的研究热点和难点,大机智能化就是其中的一个研究方向。智能列车需要解决一系列的管理和技术问题,涉及供电系统、弓网监测、车辆监测、车载网络、线路等方面,本文无法一一涉足。根据大机“开天窗”的作业方式以及大量的现场调研,论文从解决当前大机使用中迫切需要解决的问题出发,提出大机智能化的4个关键技术,即：大机信息管理技术、大机间定位及信息实时交互技术、大机司机工作状态识别技术和大机故障诊断智能决策技术。上述四个关键技术涉及人、机的管理,机器的状态监控和大机间的信息交互,此四个关键技术的合理运用,能有效地提高大机的安全性、可靠性和作业效率。论文分析了智能决策技术和方法的研究现状,探讨了信息管理技术、定位及信息交互技术、司机状态识别和故障诊断技术的研究现状。研究大机信息管理系统的功能和软件架构,设计了大机施工作业的数学描述模型和数据关系模型,并研制了数据库的主要功能。分析了大机司机工作疲劳的机理,提出了司机脸部识别和头部运动轨迹跟踪相结合的方法,实现了对司机的工作状态的实时检测和识别,采用模板匹配方法对司机的脸部进行监测,采用自适应背景的司机头部运动跟踪算法实现对司机头部运动轨迹跟踪。研制了大机司机工作状态实时识别系统并成功运用到工程现场。采用定位技术和Adhoc网络相结合实现了对大机间信息的实时交互,研究了大机间定位和信息实时交互中涉及的GPS定位技术、Adhoc网络技术和调制技术等。设计了大机间定位和信息实时交互系统的软硬件系统,在国内首次设计和研制了大机防撞预警装置。对粗糙集理论进行了分析和探讨,研究了粗糙集中进行属性约简的方法,对大机柴油机的故障数据进行预处理,同时采用最长距离法进行了属性聚类,研究了基于粗糙集的决策表属性约简算法,对大机的故障属性数据进行了约简。提出了粗糙集和案例推理相结合的方法,构建了大机故障诊断智能决策系统。研究了大机故障案例的关键技术,主要有：大机故障案例表示、大机故障案例的构建、案例相似性检索和案例修正技术。构造了大机部分常见故障案例,设计了基于粗糙集和案例推理的大机故障诊断智能决策系统总体架构,设计了大机部分故障的案例库,编写了大机故障诊断智能系统的软硬件,研制了基于案例推理的大机故障诊断智能系统。更多还原

【Abstract】 With the development of technology, especially the computer technology, microelectronics technology and semiconductor manufacturing technology have been developed rapidly, and intellectualization has been widely used in various fields, such as the intelligent mobile phone, intelligent residential district, intelligent vehicle and intelligent transport etc. The railway has been increased speed many times in large area and the mileage of high-speed train which operated in China has been to the top of the world to meet the high speed growth of the demand about railway capacity, However, the technological progress on railway maintenance is not enough that it still needs to rely on a huge crowd, especially management, operation mode and working efficiency of the Railway Maintenance Machine (hereinafter referred to as’RMM’) are often criticized. How to improve the safety, reliability and working efficiency of the RMM is the hot spot and difficulty in the current railway research, what’s more, intellectualization of the RMM is one of the research directions.There is a series of management and technical problem needed to be resolved in intelligent train, such as intelligent power supply, bow network monitoring, vehicle monitoring, vehicle network and line monitoring, etc. but not all of these subjects can be discussed in here. By a large number of field investigations and the RMM’s operation under "appointed time limit" mode, the four key problems of the RMM intellectualization have been put forward, as RMM information management technology, positioning and information real-time interactive technology between the RMMs, RMM driver working state identification technology and RMM fault diagnosis of intelligent decision technology. Those key problems relate to RMM management, condition monitoring of driver and vehicle, information interaction between RMM, so the safety, reliability and working efficiency of RMM could be improved effectively.The current research situation about intelligent decision technology and method is analyzed, furthermore, the main technology and present situation concerning intellectualization of the high-speed train is given out. In addition, the research status about information management technology, positioning and information interaction technology, driver state identification and fault diagnosis technology are analyzed in the paper.By studying the RMM information management system function and software architecture, the mathematical model about RMM construction operations and the data relation model and all software about RMM information management system database are present. The main function of the database is realized.The fatigue mechanism of the RMM driver is studied. The method of combining the driver’s face recognition and tracking head motion trajectory is proposed. The driver’s working state is detected and recognized by this way. And template matching method to monitor the driver’s face and adaptive background algorithm to realize the driver’s head motion trajectory tracking is designed. What’s more, the design and development of the real-time recognition system on RMM driver working state has applied to the railway site successfully.The real-time interaction system among the RMM information by combining the positioning technology and Adhoc is designed. It includes the positioning and information real-time interaction among the RMM which involve GPS positioning technology, Adhoc technology and modulation technology, etc. In addition, the hardware and software on positioning and information real-time interaction among the RMM is realized, and it’s the first time to design and develop early warning device for collisions in China.The rough set theory and method which using entropy for attribute reduction on rough set is discussed,and using this method to pretreat the fault data of RMM diesel engine. At the same time it processed the fault data under the method of attribute clustering. The decision-making table attribute reduction algorithm based on rough set and information entropy is studied, and the fault attribute data of RMM is reduced by this way.The intelligent decision-making system for RMM’s fault diagnosis through rough set and case reasoning is proposed. The key problems related to RMM fault cases is studied, which is including RMM fault case representation, RMM fault case construction,case similarity retrieval and case revision technology. The common fault cases about RMM and designed the framework based on rough set and case reasoning for RMM fault diagnosis of intelligent decision-making system is present. Case library about RMM fault was designed. The software and hardware relating to RMM were compiled in this paper. What’s more, the intelligent system based on case reasoning for RMM fault diagnosis is developed.更多还原