Oil plays an important role in national economy, and petroleum instrument is used to detect or acquire oil. But because now the functions of instrument are too complex, only which have work experience in this area more years can use it skillfully. Once the instrument is fault in work state, often using phone, but usually, the fault is difficult to describe, and so the work efficiency is influenced. A fault dictionary method is used to solve the problems about oil petroleum instrument and the knowledge organization and representation methods and relevant reasoning for fault diagnosis is presented in this paper. And the fault classification method uses Data Mining-method, which can easily classify the fault. And the above design method is realized using virtual instrument technology under the LabVIEW platform, and all the fault information is stored in the database, which uses the SQL server 2005 platform. Now this system uses in the field, when the instrument maintenance engineer input the fault keywords, then the information about the petroleum instrument such as fault phenomenon, solving method and fault place is disappeared in the list table. These systems uses very convenient, which are an auxiliary method to instrument maintenance engineer, and help them quickly locate the fault place.