基于粒子滤波的混杂系统故障诊断

发布时间：2018-04-08 12:10

本文选题：混杂系统故障诊断　切入点：粒子滤波　出处：《北京交通大学》2017年硕士论文

【摘要】：混杂系统是指既包含连续动态特性,又包含离散动态特性,而且两种动态特性相互作用的系统。一般规模比较大的自动控制系统都具有混杂特性,导致系统复杂并且可能出现多种多样的故障,为了保证系统的正常运行和安全问题,必须有高效准确的故障诊断方法;对于混杂系统这样的非线性系统,粒子滤波算法比卡尔曼滤波算法更具优势。本文研究了基于粒子滤波算法的混杂系统故障诊断问题,具体切入点为模型建立方案、混杂系统故障分类和具体故障诊断方法及其性能评定,对非线性系统故障诊断的方法进行了探究。本文所做主要工作如下:(1)分析了粒子滤波算法的粒子退化问题和重要性密度函数选择问题,比较算法SIR和UPF的优缺点,作为理论基础。针对经典的系统故障模型逻辑简单、故障人工注入单一且不具备随机性的特点,根据随机混杂系统理论,建立了随机混杂自动机(SHA)模型,模拟了更为复杂多样的故障。(2)建立了水箱水位控制系统的随机混杂自动机模型,对该系统出现的故障进行抽象、分类,总结为执行故障、观测故障、切换故障,并针对常用的模态—变量混杂系统粒子滤波估计方法难于应用在复杂逻辑模型上的特点,提出了基于粒子滤波的混杂系统切换统计故障诊断方法,并进行了验证。(3)为了能使模型表达能力更强,灵活度更高,本文引入了分层混杂系统理论,在理论的基础上给出实现分层混杂系统理论的具体建模方法,提出了上层模态和子模态的概念,且在具体建模过程中发现上层模态切换出现断层问题,提出中转模态概念和方法进行解决,完成了分层混杂自动机建模。(4)建立了混杂系统受控迁移伴随故障模型,利用SIR粒子滤波方法和平滑处理的思想方法对分层混杂系统的上层模态切换故障进行了故障诊断方法方案研究,并对其性能指标进行分析对比。采用CTCS-3级列车运行控制系统等级转换模型进行了说明验证。
[Abstract]:Hybrid systems are systems that contain both continuous and discrete dynamic characteristics and interact with each other.Generally large automatic control systems have hybrid characteristics which lead to the complexity of the system and the possibility of a variety of faults. In order to ensure the normal operation and safety of the system there must be an efficient and accurate fault diagnosis method.For nonlinear systems such as hybrid systems, particle filter algorithm has more advantages than Kalman filter algorithm.In this paper, the problem of hybrid system fault diagnosis based on particle filter algorithm is studied. The starting point is modeling scheme, fault classification, fault diagnosis method and performance evaluation of hybrid system.The method of fault diagnosis for nonlinear system is discussed.The main work of this paper is as follows: 1) the particle degradation problem and the importance density function selection problem of particle filter algorithm are analyzed. The advantages and disadvantages of SIR and UPF are compared as the theoretical basis.In view of the simple logic of the classical system fault model, the single fault artificial injection and the lack of randomness, a stochastic hybrid automaton (SHA) model is established according to the stochastic hybrid system theory.The random hybrid automata model of water level control system of water tank is established by simulating more complex and varied faults. The faults of the system are abstracted, classified, summarized as execution faults, observed faults, and switched faults.Aiming at the difficulty of particle filter estimation in complex logic model, a new statistical fault diagnosis method for hybrid system switching based on particle filter is proposed.In order to make the model more expressive and flexible, the theory of hierarchical hybrid system is introduced in this paper. On the basis of the theory, a concrete modeling method is given to realize the theory of hierarchical hybrid system.The concepts of upper mode and sub mode are proposed, and the fault problem of upper modal switching is found in the process of modeling. The concept and method of transfer mode are proposed to solve the problem.The hierarchical hybrid automaton modeling. 4) the controlled migration associated fault model of hybrid system is established.In this paper, SIR particle filter method and smoothing method are used to study the fault diagnosis scheme of upper modal switching fault of hierarchical hybrid system, and its performance index is analyzed and compared.The class conversion model of CTCS-3 train operation control system is used to explain and verify.
【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】：TN713;TP273

【引证文献】