Based on the characteristics of an
expert system and its possible applications to the air pumping process of a large-scale
vacuum metallurgical system, the difficulties in trouble-shooting of the
pumping system are discussed in detail to make sure the possibility and practicality of introducing an expert system into the pumping system so as to ascertain what the expert system has to do. With the RH-KTB vacuum refining system of BAOSTEEL in Shanghai taken as instance, how its expert system was developed and works are analyzed including the operation monitoring mode and its knowledge base. The software programmed for the expert system can provide not only the real-time assessment and fault forecast during the operation of the vacuum refining system but finding out and displaying the possible causes of a fault by way of reverse
reasoning from its related objective if a fault is occuring. In particular the reasoning efficiency and precision of the
Expert system are improved further through self-learning the thresholds of monitoring parameters. The results of off-line tests and on-line applications both proved that the expert system attains the anticipated goal and is available to utilize wider.