By designing the prime frame of equipment fault diagnosis system and using equipment fault records of large screen system as a sample data, this article demonstrates and analyzes the application of Rough Set in the equipment fault diagnosis system. The following work has been done: by the method of combining the Rough Set theory with the Statistical theory, the datasets with missing values of records can be recovered; the algorithm of knowledge acquisition of Rough Set based on Granular Computing can be used to transform the attribute in decision sheet into matrix objects, then the complex knowledge simplification is transformed into matrix operation, thus the time consuming can be deducted, and the knowledge of fault diagnosis can be acquired automatically and more effectively.