Aiming at the defects of subjectivity, limitation and singleness of FMECA in risk analysis of complex system, an improved method of FMECA based on data mining is proposed. The corresponding algorithm is optimized by using Python and database through data mining FMECA risk list. The risk level assessment strip diagram, fault hierarchy diagram and improved risk matrix diagram are given combined with the case analysis. The results show that the improved FMECA can make full use of relevant data, making the analysis results more rapid, accurate and comprehensive, helping to further discover the relationship between potential failures of the equipment, and providing support for the intelligent health management of the equipment.