Abstract: The extension data mining methods and pattern recognition technology are combined to solve the problem of fault diagnosis involved with large amounts of data. By taking advantage of the theory of rough set to make attribute reduction and weight value determination, it ensures the objectivity and stability of weights. Then it verifies the reliability of the model with the example of fault diagnosis of automobile engine. Finally, according to the classification method of extension set based on the transform, this paper presents a method for fault prevention.

Key words: Extenics; data mining; fault diagnosis; rough set