关键词: 故障诊断, 定量推理, 人工智能

Abstract:

A new diagnosis method based on intelligent information processing is described and analyzed, which can deal with complicated combination patterns and temporal characteristics of the faults in complex dynamic systems. By quantifying the reasoning mechanism and optimizing the reasoning environments, it is improved further in diagnosis capability, real time performance and application scope. The reasoning mechanism is quantified by introducing information of probability. Because here a more practical assumption is used that new faults can take place during the diagnosis process, a new mechanism called′ candidates extending′ is introduced to describe this phenomenon, and some formulae are provided for probability evaluation. By quantifying the reasoning mechanism, the diagnosis results can be measured quantiatively and the diagnosis process can be controlled accurately. Optimizing the reasoning environments is realized by introducing information of entropy, which can improve diagnosis effectiveness greatly.

Key words: fault diagnosis, quantiative reasoning, a rtificial intelligence