Abstract: With the increase of the nuclear power plant expert system knowledge base, the consistency checking of knowledge base becomes difficult, and the diagnostic reasoning efficiency of the system also decreases. To solve this problem, fuzzy Petri net was used in knowledge representation and reasoning of the expert system. Fuzzy Petri net analysis techniques was used in calibration and maintenance of the knowledge base, and the diagnostic reasoning was carried out based on diagnostic rules and operating mechanism of Petri net. Simulation results show that the fuzzy Petri net is used in expert system, expert system can effectively carry out the knowledge base consistency test and fault diagnostic reasoning.