基于神经网络的计算机联锁系统寿命评估方法

发布时间：2018-04-16 17:21

本文选题：计算机联锁系统 + 寿命预测　；参考：《北京交通大学》2017年硕士论文

【摘要】：计算机联锁系统是铁路信号的核心技术装备,是保证安全高效运行及铁路车站作业的重要环节。现阶段,我国铁路计算机联锁系统的寿命周期管理办法主要参考沿用传统继电联锁设备的有关规定,缺乏针对系统中电子设备使用寿命的分析与评估方法,这在很大程度上限制了对计算机联锁系统使用寿命的科学管理。本文从系统的体系结构出发研究了一种基于神经网络的计算机联锁系统寿命评估方法。主要工作如下:(1)综合分析了计算机联锁系统应用现状和各领域系统寿命评估方法研究的现状,提出了适合于计算机联锁系统的寿命评估方案,给出了寿命评估的具体实施步骤;(2)结合联锁系统硬件结构和部件功能分析,以二乘二取二制式联锁系统为主要研究对象,构建了联锁系统故障树模型;通过定性分析得出故障树模型的最小割集,建立了部件故障与系统失效之间的关系,构造了神经网络训练数据集;(3)研究了神经网络预测性能提升方法,利用粒子群算法对GRNN神经网络进行中心神经元宽度矩阵的参数寻优,有效提高了神经网络预测的精度;(4)为了充分体现联锁系统硬件冗余结构特点,基于GRNN神经网络、改进型GRNN神经网络和BP神经网络,建立了三种不同的计算机联锁系统寿命评估模型;(5)对三种基于神经网络方法的联锁系统寿命评估模型,进行了性能对比分析,并结合系统特点,给出了网络性能最优的评估模型。最后,以我国铁路广泛使用的AB型联锁系统为对象,利用现场运营数据,进行系统寿命评估,检验本文提出评估模型和方法的有效性。本文通过理论分析和实例验证,给出了一种基于神经网络的计算机联锁系统寿命评估方法,以实现对计算机联锁系统服役寿命的科学预测评估,可为我国计算机联锁系统的运营管理提供借鉴和参考。
[Abstract]:Computer interlocking system is the core technology equipment of railway signal and an important link to ensure safe and efficient operation and railway station operation.At present, the life cycle management method of railway computer interlocking system in our country mainly refers to the relevant provisions of traditional relay interlocking equipment, and lacks the analysis and evaluation method for the service life of electronic equipment in the system.This limits the scientific management of the life of computer interlocking system to a great extent.In this paper, a method of evaluating the life of computer interlocking system based on neural network is studied based on the architecture of the system.The main work is as follows: (1) A comprehensive analysis of the application status of computer interlocking system and the present situation of life evaluation methods in various fields are given, and a life evaluation scheme suitable for computer interlocking system is put forward.Based on the analysis of hardware structure and component function of the interlocking system, the fault tree model of the interlocking system is constructed by taking the two-plus-two-mode interlocking system as the main research object.Through qualitative analysis, the minimum cut set of fault tree model is obtained, the relationship between component failure and system failure is established, and the neural network training data set is constructed.Particle swarm optimization algorithm is used to optimize the parameters of central neuron width matrix of GRNN neural network, which effectively improves the precision of neural network prediction. In order to fully reflect the characteristics of hardware redundancy structure of interlocking system, it is based on GRNN neural network.Improved GRNN neural network and BP neural network, three different computer interlocking system life evaluation models are established. The performances of three interlocking system life evaluation models based on neural network are compared and analyzed.Combined with the characteristics of the system, the optimal evaluation model of network performance is given.Finally, taking AB type interlocking system, which is widely used in railway in China, as the object, using field operation data, the system life evaluation is carried out, and the validity of the evaluation model and method proposed in this paper is verified.In this paper, a method of evaluating the service life of computer interlocking system based on neural network is presented through theoretical analysis and example verification, so as to realize the scientific prediction and evaluation of the service life of computer interlocking system.It can provide reference and reference for the operation and management of computer interlocking system in China.
【学位授予单位】：北京交通大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：U284.362;TP183

【参考文献】