Abstract: Based on the data mining technology, the pre-processing method of intelligent root selection data for nuclear-level pipeline hanger and support is studied. The classification method of root selection pre-processing is studied. The data pre-processing flow is designed and the priority order of root selection of hanger and support is determined; Based on SOM clustering algorithm, the calculation process of intelligent selection data of the root of hanger and support is studied, the experimental platform is designed. Based on the actual engineering data, the feasibility and the effectiveness of the algorithm are verified, the data pre-processing is proved and the clustering effect is obvious.