面向原发性因素溯源的汽轮机故障系统级建模理论研究

The existing researches on fault diagnosis of steam turbine, is mainly aiming at some key components and parts such as rotor-bearings system, etc., and mainly focused on the diagnosis of some typical fault patterns. Usually, component-level fault diagnosis can only recognize the induced fault, without further traceability and cure on their underlying factors. From the systemic point of view, steam turbine fault shows obvious characteristics as atypical predisposing factors, multi-factor coupling and dynamic evolution, which put forward the urgent request to system-level fault diagnosis. However, its researches are still not enough at present, and need to be further strengthened. This project is based on the power plant steam turbine as the research objective, on system-level modeling of steam turbine fault as the research thread, on the tracing back to the underlying failure factors as research goal, on establishment of a complete representation hierarchy on system state of multi-dimensional, multi-parametric and multi physical quantities as the research breakthrough point, on the combined layering decomposition on structure and function of system as research method. Using a comprehensive theoretical and experimental simulation strategy, both multi-factor coupling mechanism and evolution law on steam turbine fault are deeply explored and studied. Finally, a series of systemic solutions to underlying factors traceability are proposed, including the system-level modeling technique for steam turbine fault, the algorithm for decoupling control on coupled factors and the inverse approach based on inverse system analysis. Thus, theoretical and technical support is supplied to improve the level of system-level fault diagnosis of steam turbine.

现有的汽轮机故障诊断研究，主要针对的是设备中某些关键的零部件如转子-轴承系统等，并主要集中于典型故障模式的诊断。零部件级故障诊断往往只能诊断出诱发性故障，无法进一步追溯故障的发生根源并对其进行根治。从系统角度看，汽轮机故障具有明显的诱发因素非典型性、多因素耦合性以及动态演变性，这对系统级故障诊断提出了迫切的要求，但是目前这方面的研究还不够，有待进一步强化。本项目以电厂汽轮机为研究对象，以汽轮机故障的系统级建模为主线，以追溯故障的发生根源为目标，以建立系统状态的多维、多参数、多物理量完备描述体系为切入点，以系统的结构-功能层次化协同分解为手段，采取理论仿真与试验模拟综合的研究策略，深入探索汽轮机故障的多因素耦合机理，研究故障的动态演变规律，提出一套面向原发性因素溯源的汽轮机故障系统级建模技术、耦合因素的解耦控制算法和基于逆系统分析的反求方法，为提高汽轮机系统级故障诊断水平提供理论和技术支撑。

现有的汽轮机故障诊断研究，主要针对的是设备中某些关键的零部件如转子-轴承系统等，并主要集中于典型故障模式的诊断。零部件级故障诊断往往只能诊断出诱发性故障，无法进一步追溯故障的发生根源并对其进行根治。从系统角度看，汽轮机故障具有明显的诱发因素非典型性、多因素耦合性以及动态演变性，这对系统级故障诊断提出了迫切的要求，但是目前这方面的研究还不够，有待进一步强化。本项目以电厂汽轮机为研究对象，主要在汽轮机系统状态的描述体系、汽轮机故障的系统级建模方法、汽轮机系统故障原发性因素的追溯技术以及原型系统开发与实证等几个方面开展研究，建立了系统状态的多维、多参数、多物理量完备描述体系，对汽轮机故障的多因素耦合机理与动态演变规律有了深入认识，进而提出系统的面向原发性因素溯源的汽轮机故障系统级建模技术、耦合因素的解耦控制算法和基于逆系统分析的反求方法，并构建了原型软件系统。所取得的系列研究成果，为提高汽轮机系统级故障诊断水平提供了理论和技术支撑。