Taking the fire control system of some equiment as an example, combining text classification technique and fault diagnos method based support vector machines, through founding fault feature thesaurus, generating fault vector thesaurus by adopting Boolean Models, trainning and evaluating fault of fire control system for this equipment by SVM algorithm, the better test result was acquired, whose fault recognition rate reached up to 70%. For equipment maintenance, especially for battlefield repair, it was very significant that making fault diagnosis for equipment with this method. By this method, maintenance personals was extricated themselves from cockamamiecheck with instrument, so that fault ispecting points were confirmed in faults thesaurus rapidly and conveniencely, and rush repairs for equipment would come true, which would create favorable conditions for snatching the initiative on the battlefield.