Abstract: The discovery of fault information is conducive to improve the quality of fault diagnosis and the efficiency of fault diagnosis.Since fault information presented in the form of rough-up text description,in order to avoid waste of information resources,to make use of fault information fully and effectively,according to Dealer Management System(DMS) data of pickup,fault information mining machinery is built by production representation method based on semantic description.It is used to analyze fault information.Uncertainty reasoning of credibility is used to infer the parts that cause trouble and to verify the reliability of rule conclusion.

Key words: fault, association rule, data mining, production rule