中文摘要：

为解决大规模复杂系统故障诊断中计算复杂性高的问题,提出一种基于Petri网的在线故障诊断方法.首先,建立诊断对象的规范Petri网模型;其次,提出模型的严格最小库所不变量和特征库所不变量集合,并借助特征库所不变量集合描述Petri网模型的结构信息;最后,基于特征库所不变量集合提出任意当前标识的故障函数,并利用故障诊断函数完成故障识别和定位.结果表明：该故障诊断方法采用了系统结构信息,无需遍历系统状态空间,具有多项式级的计算复杂性,能够满足实时性要求.

英文摘要：

For fault diagnosis in large complex systems, a on-line fault diagnose method is proposed to solve the problem of high computational complexity. First, modeled a Petri net model. Secondly, proposed the strict minimal place-invariant and the set of characteristic plaee-invariant, so that might describe the structure information of Petri net model. Finally, based on the set of characteristic plaee-invariants, the failure function for any current marking is proposed. And then, utilized this failure function to diagnose and locate the faults. The result shows that this fault diagnosis method with the structure information dose not need traverse all states space of system. Furthermore, this method is with the computa- tional complexity of polynomial, which makes this method meet the real time requirements.