Stage feature extraction of flight data based on clustering method

It is an important direction for the development of aircraft health management system to utilize flight data to conduct fault diagnosis and trend prediction. The status of fault is usually correlated with specific flight stage, but flight data recorded by the recorder usually have no corresponding stage feature, which presents great challenges to the application of fault diagnosis system. This paper presents the idea to apply clustering technologies to the stage feature extraction of flight data according to the characteristics of flight data and the advantages of clustering technologies. Based on the idea, this paper conducts research and optimization design work for kernel k-means and utilizes authentic flight parameters to develop simulation analysis work, and the simulation results shows that this idea can solve the stage feature extraction problem of flight data.