Abstract: The volume of Electric Multiple Units (EMU) fault data is becoming larger and larger, showing the characteristics of geometric growth, and the sources are becoming more and more extensive. However, the accumulation of fault data has not significantly improved the level of fault analysis. In order to study the safety regular pattern of EMU and dig out the data value of massive historical faults, based on the EMU safety big data platform, using multi-source data collection, fusion processing, and big data machine learning technology, this paper designed the EMU safety law analysis system architecture and functions. At present, the system has been researched and developed, and the safety regular pattern of CRH380 series EMU have been verified. The construction of the system is conducive to improving the safety management level of the EMU, providing effective decision-making suggestions for the technical management personnel of the EMU, and reducing maintenance costs.