Abstract: This paper focused on solving the problem of EMU association rule mining under big data, proposed an improved T-MR-Apriori algorithm for association rule mining of big data based on Apriori algorithm. The improved algorithm combined Hadoop technology, performed two times MR distributed computing process, completed the whole process of association rule mining, improved the efficiency and accuracy of association rule mining under massive data. The actual EMU operation and maintenance data were used to verify the algorithm, which proved that the algorithm had good speed of mining in mass data and cannot reduce the performance of mining. The method would be applied to data mining and visualization of traction motor operation and maintenance in EMU.