动态系统故障机理的混杂特征及其建模方法

Failure mechanisms of the dynamic system manifest typical hybrid characteristic for the reason that its failure process is comprehensively influenced by the internal and external factors, component failures and the coupling among them. By taking the typical dynamic system as the research object, firstly, a full understanding on the hybrid characteristic of failure mechanisms of the dynamic system is obtained through the coupled impact of the discrete events, continuous parameters and coupling among them on the dynamic system. Secondly, the stochastic hybrid Petri net method is developed considering the features of the reliability discipline to construct the hybrid failure mechanism model for dynamic systems. In the proposed method, the discrete event model and the continuous dynamic model are integrated and the dynamic interaction between them is described by the dynamic transition mechanisms (immediate transition, linear transition and integral transition). Then random factors injecting method is promoted to obtain all the revolution paths by simulation. Lastly, an adaptive step model driving algorithm is proposed to deal with the random jump, as well as to realize the sensitive detection and precise positioning of the deterministic events and random events. And the efficiency of the simulation would be improved with the assurance that no events misses. The hybrid failure mechanism and its modeling method would lay the foundation of the failure cause analysis, quantitative reliability design and fault diagnosis of dynamic systems.

动态系统的故障机理具有明显的混杂特性，其故障演化过程受到内外因素、单元故障及其耦合作用的综合影响。首先，以典型动态系统为研究对象，从离散事件、连续参数及其耦合作用对动态系统故障的耦合影响入手，形成对动态系统故障机理混杂特征较为完整的认识。然后，结合可靠性领域特点提出随机混杂Petri网方法，集成离散事件模型和连续动态模型，通过动态变迁机制（瞬时变迁、线性变迁、积分变迁）描述模型间的动态交互，并在此基础上研究随机性因素注入方法，通过仿真输出各种可能的故障演化路径。最后，提出一种能处理随机性跳变的自适应步长模型驱动算法，实现对随机故障事件的敏感检测和精确定位，在保证不发生事件漏检的前提下提高仿真效率。混杂故障机理及其建模方法可为动态系统科学的开展故障原因分析、可靠性定量设计和故障诊断奠定基础。

动态系统的故障传播过程是由离散事件、连续特性及其相互作用共同驱动的，具有显著的混杂特征，为故障规律认知与建模带来了较大的难度。首先，从离散事件、连续参数及其耦合作用对动态系统故障的耦合影响入手，分析出系统故障过程中存在参数连续退化，系统连续特性和单元离散故障之间的耦合作用，给出了离散与连续双维度下的动态系统故障混杂传播的定义，分析了其混杂影响要素以及混杂传播特征，形成对动态系统故障机理混杂特征较为完整的认识。然后，结合可靠性领域特点提出随机混杂自动机（Stochastic Hybrid Automata, SHA）方法，集成离散事件模型和连续动态模型，通过动态变迁机制（瞬时变迁、线性变迁、积分变迁）描述模型间的动态交互，在SHA基础上引入状态内部的随机性，考虑噪声、结构参数的随机退化及使用环境等不确定因素对系统性能的影响。随后，提出一种能处理随机性跳变的自适应步长模型驱动算法，将随机性事件函数和确定性事件函数一并纳入到事件条件中，通过分层处理方法，实现对随机故障事件的敏感检测和精确定位，在保证不发生事件漏检的前提下提高仿真效率。最后，以电加热炉温度控制系统为例，对本文提出的方法进行了验证。项目首次将混杂理论应用到可靠性建模领域，能够描述离散单元故障和连续过程参数之间的交互作用，为动态系统故障规律认知、故障原因分析和故障诊断提供参考。