基于主成分分析法与贝叶斯网络的汽轮机故障诊断方法  

Steam Turbine Fault Diagnosis Methods Based on the Main Constituent Analysis Method and Bayesian Network

作　　者：韩璞[1] 张德利[1] 韩晓娟[1] 焦嵩鸣[1] 

机构地区：[1]华北电力大学控制科学与工程学院,河北保定071003

出　　处：《热能动力工程》2008年第3期244-247,共4页Journal of Engineering for Thermal Energy and Power

摘　　要：在利用贝叶斯网络进行汽轮机故障诊断时,汽轮机故障诊断模型的建立直接影响着诊断过程的复杂程度,因此建立贝叶斯网络模型是首先要解决的问题,反映故障状态的特征参数提取是建立模型的重要环节。通过对主成分分析方法提取故障特征的讨论,提出了基于主成分分析方法和贝叶斯网络的汽轮机故障诊断模型建立方法,并与传统的频率特征建模方法进行了比较。结果表明,应用主成分分析法和贝叶斯网络建立的汽轮机故障诊断模型简洁,易于推理,提高了汽轮机故障诊断的效率。When Bayesian network is used to diagnose a fault,the establishment of a model for fault diagnosis of steam turbines has a direct bearing on the complexity of the fault diagnosis process.Therefore,to establish a model of Bayesian network becomes an issue of first priority and the collection of characteristic parameters reflecting the fault status constitutes an important link for setting up a model.Through a discussion of the collection of fault characteristics by using the main constituent analysis method,presented was the modeling method for steam turbine fault diagnosis based on the main constituent analysis and Bayesian network.In addition,the proposed method has been compared with the traditional frequency characteristics modeling method.The results show that the model in question for turbine fault diagnosis is simple and lends itself to easy reasoning,thus enhancing the efficiency of turbine fault diagnosis.

关 键 词：汽轮机 故障诊断 主成分分析法 贝叶斯网络 

分 类 号：TK268[动力工程及工程热物理—动力机械及工程]