基于粗糙集理论的汽轮机组回热系统故障诊断模型

Fault Diagnosis Model for the Regenerative Heating System of a Steam Turbine Unit Based on a Rough Set Theory

在分析了汽轮机组回热系统现有故障诊断方法无法解决冗余征兆的不足之后 ,提出了一种基于粗糙集理论的故障诊断模型。该模型从回热系统典型故障模式出发 ,通过连续征兆属性的离散化建立了故障诊断决策表 ;利用遗传算法实现了故障征兆属性约简 ,并提出了结合领域知识的最小约简择优策略 ,然后通过给出的决策规则约简的基本原则 ,得到用于故障诊断的决策规则库。在应用该模型进行故障诊断时 ,用待诊实例的离散化了的故障征兆属性与规则库中的诊断决策规则进行匹配 ,对返回的诊断决策规则进行综合评价 ,并得出诊断结论。利用电站仿真机模拟典型故障进行了故障诊断模型的验证 ,实践表明 ,该模型可以有效地约简冗余的故障征兆 ,并具有较好的诊断效果和一定的容错能力。

After an analysis of the insufficiency of current fault diagnostic methods used for the regenerative heating system of a steam turbine to resolve the problem of redundant fault symptoms the authors have proposed a new fault diagnosis model based on a rough set theory. With the typical fault modes of a regenerating heating system being taken into account a fault diagnostic decision table was established through a discretization of continuous fault symptom attributes. A reduction of the fault symptom attributes was realized by making use of a genetic algorithm. An optimal selection stratagem of minimal reduction is proposed based on domain knowledge. Then, a decision rules base for fault diagnosis was set up through the basic principle of decreasing the given decision rules. When the proposed model is employed for fault diagnosis the discretized fault symptom attributes to be diagnosed are first matched with the diagnostic decision rules in the rules base. The returned diagnostic decision rules will undergo a comprehensive evaluation with a diagnostic conclusion being reached. The simulation of typical faults by a power plant simulator was performed to verify the fault diagnosis model. Engineering practice shows that the proposed model is highly effective in reducing redundant fault symptoms and credited with a good fault-diagnosis effect as well as a fair fault-tolerant capability.