Application of Adaptive Mathematical Models to a T-37 Pilot Performance Measurement Problem
MELPAR FALLS CHURCH VA FALLS CHURCH
Pagination or Media Count:
The report documents experimental research on a new method of deriving performance measures and criteria for use in automated pilot performance evaluation. Data recorded on board a T-37B aircraft tail number 58-1948 were submitted to a previously implemented system of adaptive mathematical models AMM. The results were analyzed to determine the practical capability of the AMM in automatically deriving measures and criteria. Flight data for a series of performances of the Lazy 8 and Barrel Roll maneuvers were processed first by a set of Boolean functions. These functions describe the data in the form of Boolean time sequences BTS, which are then operated upon by the AMM to derive three types of performance measures 1 State Transfer Measures, which are based on overall trends in the performance 2 Absolute Measures, which are based on a comparison of actual performance with some reference and 3 Relative Measures, which are based on relations among performance variables. The results show that the AMM system can be used to effect a systematic attack on the problems of performance measurement using representative flight data. Face-validity of measures derived by the AMM is illustrated by comparison with performance evaluations made by an instructor pilot.