基于趋势分析和SDG模型的故障诊断  

Fault diagnosis based on trend analysis and SDG

作　　者：刘敏华[1] 萧德云[1] 

机构地区：[1]清华大学自动化系,北京100084

出　　处：《控制理论与应用》2006年第2期306-310,共5页Control Theory & Applications

基　　金：国家高技术研究发展计划(863计划)资助项目(2003AA421310)

摘　　要：基于趋势分析的故障诊断受数据影响较大,基于定向符号图(signed d irected graph)模型的故障诊断难于建模和推理,而两者结合以提高故障诊断效果的方法还未被考虑.本文将趋势分析的趋势信息和SDG模型的节点信息进行了比较,阐明了两者信息互补的特点,然后利用趋势分析建立SDG模型,并利用SDG模型改进基于趋势分析的故障诊断,建立了数据驱动方法和基于模型方法相结合的故障诊断方法.CSTR(continuous stirred-tankreactor)实例分析表明,基于趋势分析和SDG模型的故障诊断方法提高了诊断的准确性和精确性.Fault diagnosis based on trend analysis is affected by data, while the one based on signed directed graph （SDG） is difficult to model and infer. The combination of both in order to improve the effect of fault diagnosis has not been considered yet. The trend information of trend analysis is compared with the node information of SDG, which shows that they complement each other. Then trend analysis is applied to SDG modeling, and the fault diagnosis based on trend analysis is modified with corresponding SDG. So the new data-driven and model-based method of fault diagnosis is built. The example of continuous stirred-tank reactor（CSTR） illustrates the veracity and accuracy of fault diagnosis based on trend analysis and SDG.

关 键 词：故障诊断 定向符号图 趋势分析 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]