Criterion Setting for Objective, Fourier Analysis Based Pilot Performance Metrics

Abstract

This study reports the development and evaluation of time series based objective pilot performance metrics. From a previously developed array of autocorrelation and Fourier analysis based metrics, five Fourier-based metrics that employed a threshold value were chosen to investigate their effectiveness in separating pilots who, based on instructor pilot (IP) evaluations, had either passed or failed a particular segment of an instrument proficiency check flight. An instrument landing system (ILS) approach was chosen for analysis based on IP feedback of what flight segments were most difficult to evaluate, had greatest sensitivity to overall pilot performance, and greatest criticality to the flying task. Further analysis showed that an optimal value for the criterion value could be found that most effectively separated those pilots that had passed the ILS segment from those who had failed. Criterion setting methods without external criteria using multidimensional scaling and cluster analysis techniques are also discussed.