基于半定量SDG模型的化工过程故障诊断

Chemical process fault diagnosis using semi-quantitative SDG model

符号有向图(SDG)能够直观地反映系统的复杂因果关系,可以看出故障传播的路径,但是定性SDG故障诊断模型缺少变量之间的定量信息,推理过程中会产生虚假解,分辨率不高。该文针对SDG的不足,引入模糊隶属度,形成半定量SDG模型,即模糊SDG模型。采用反向推理和正向验证相结合的混合推理方法,并把结论编写为"If-Then"形式的知识规则,以便于利用专家系统对系统进行在线诊断。

Signed directed graph （SDG） can directly reflect complex causal relationships of systems with the fault propagation path being observed from the SDG. However, for less quantitative information, pure qualitative SDC- model always has many false solutions in the reasoning process with low resolutions. The fuzzy membership was combined with SDG to form a semi-quantitative model. Fault diagnoses were carried out for the system using the positive and negative reasoning of the system to come to the conclusions. The conclusions are written into knowledge rules as ＂If Then＂ form to facilitate online diagnosis using expert systems.