Determining Most Likely Flight Profiles from Aircraft Usage Data for Damage Prognosis

Military aircraft experience multiple types of fatigue damage caused by various factors, including thermal effects and structural loads. Damage prognosis for such aircraft requires more comprehensive usage information than traditional structural loading spectra can provide. The desired usage data predictions should be sufficient to reconstruct flight environments and will consist of multidimensional, time-varying, correlated flight parameters. Existing techniques for fatigue loading spectra fail to model and generate such data due to the complex correlations between flight parameters and the associated uncertainty. In this work, three probabilistic approaches are proposed to identify the most likely flight profiles from past aircraft usage data. The identification procedures in these approaches are developed by addressing three different aspects of usage data, which are variations in main trends of flight data, variability in pilot control, and cumulative damage. These approaches were compared and demonstrated through their applications to the usage data collected from a set of Touch-and-Go’s.