【作者】 孙可；

【导师】 孙斌； 金心宇；

【作者基本信息】 浙江大学 ， 电路与系统， 2012， 硕士

【摘要】 近年来,中国的高速铁路飞速发展,特别是京沪高铁的运营使用标志着中国的高铁技术已领先于其它各国。当高速列车的速度在300km/h以上时,通常需要采用CTCS-3( China Train Control System Level Three)专用的网络—GSM-R(GSM for Railway)作为列控信号的通信平台。由于高速铁路沿线的地形环境十分复杂,因此影响网络正常通信的因素千变万化,其中干扰现象是最为普遍,也是影响程度最深的因素之一。严重的干扰甚至会造成通信中断,引起重大的事故,所以对干扰现象进行深入研究,并提出对其优化的策略具有至关重要的意义。在对干扰现象的研究过程中,干扰的检测、定位和优化是当前业界研究的重点内容。本文阐述了干扰检测的三种思路：TCH(业务信道)性能测量、切换性能测量和掉话分析,并详细介绍了定位和排查的常用方式,在此基础上对实测数据进行了分析。同时,本文采用NBtree算法和J48算法对海量实测数据进行了干扰事件的分类挖掘,其结果表明实测路段的干扰现象较为严重,并且呈现不均匀分布；此后本文使用常用的干扰定位方法进行排查,发现干扰现象频繁发生往往是由于相应路段的频率分配不满足干扰约束条件所造成的。针对此种情况,本文从频率规划的角度出发提出了一种新的频率分配模型,该分配模型主要从干扰约束和频率利用率两方面进行考虑；然后在此模型的基础上提出了基于遗传禁忌算法的干扰优化策略,该优化策略以干扰限制条件和频率利用率为目标函数来得到合理的频率规划方案,以此减少系统内干扰的发生,从而提高了GSM-R网络的通信性能。本文从电磁波传播的角度出发研究了传统干扰优化方法的优劣,发现传统方法往往通过调整场强覆盖来减少干扰,比较被动而且可操作性较差；然后基于频率分配模型仿真对比了基于遗传算法的干扰优化策略和本文所提出的优化策略,其结果表明基于遗传禁忌算法的干扰优化策略克服了遗传算法自身的缺陷,杜绝了“早熟”现象,而且该优化策略的收敛性要远远强于基于遗传算法的经典优化策略。因此,与以往的干扰优化方法相比,本文所提出的优化策略具有可操作性强、性能好、收敛性高等特点。更多还原

【Abstract】 During the last decade, the high-speed railway has developed rapidly in China. Especially, the opening of Beijing-Shanghai high-speed railway indicates that China has been ahead of the rest world in the technology of high-speed railway. When the speed of high-speed railway exceeds 300 kilometer per hour, GSM-R (GSM for Railway) is generally utilized, which is the CTCS-3’s (China Train Control System Level Three) private network. Because the terrain environment along the high-speed railway is rather complex, there are many kinds of factors that affect the communication network, thereinto, the interference phenomenon is the most common and influential, and serious interference even causes communication interrupt which leads to major incidents. Hence, delving into the interference phenomenon of GSM-R network and proposing the optimization strategy is of importance.In the research of interference phenomenon, the detection and optimization of interference are key subjects of current industry. This paper expounds three methods to detect interference and introduces the common ways to locate interference. At the same time, this paper utilizes NBtree algorithm and J48 algorithm to classify mass measured data based on interference event, and its result indicates that the interference phenomenon is rather serious along the testing section and has the feature of nonuniform distribution. After checking this problem, it is discovered that the improper frequency allocation leads to the frequent occurrence of interference. As to this condition, this paper proposes a new frequency planning model. This model takes account of the following two aspects, interference constraints and the utilization of frequency. And then this paper proposes a new interference optimization strategy based on Generic-Tabu Algorithm by using the proposed model, and this strategy sets the constrain condition of interference and utilization of frequency as the objective function in order to get the reasonable frequency planning scheme, which avoids the interference in GSM-R network and improves the performance of the whole network.In the process of analyzing the traditional interference optimization methods, this paper studies the traditional optimization methods’advantages and disadvantages in the view of the electromagnetic wave propagation. Its result shows that the traditional methods reduce interference by adjusting the covering of field strength, so this kind of methods is rather passive and not easy to implement. By comparing the interference optimization strategy based on Genetic Algorithm to the strategy proposed by this paper via simulation, the result shows that the strategy proposed by this paper overcomes the defect of Genetic Algorithm. Moreover, its convergence is much better than the optimization strategy based on Genetic Algorithm. Therefore, the optimization strategy proposed by this paper is much fitter for GSM-R network.更多还原