Abstract: Complex fault mechanism and operation parameters of the tunnel boring machine (TBM) were analyzed， and the method of rough set and decision tree algorithm applying to data mining was studied. Take several MATLAB 7.0 dispersed data of tunnel boring machine cutter head as an example， the redundancy attribute of fault samples was reduced by the combination with the rough set attribute reduction algorithm. The rules were extracted with the decision-making tree algorithm. The C4.5 algorithm and the improved C4.5 algorithm were implemented with the data mining tool Clementine， with the results compared. The data was tested by the VB programming. The results showed that the fusion algorithm is a rapid， effective and reliable approach for fault detection and diagnosis.

Key words: tunnel boring machine, data mining, rough set, decision tree, fusion algorithm