Deriving Aircraft Characteristics from Arbitrary Flight Test Data

Despite advancing in technology, computing power, and the ability to integrate instrumentation with aircraft over the last half century, the United States Air Force has never meaningfully changed the way flight test is conducted over that time. The precision required to accomplish desired maneuvers often results in multiple failed attempts, resulting a significant waste of time, money, and flight hours in critical programs. A properly instrumented aircraft can be flown in an arbitrary manner, sweeping through or near test points, and create the same results that exactly measuring the original test point would provide. By using linear models and an algorithm based on Gaussian Process Regression, continuously-collected data was used to predict short period damping ratio and damped frequency. Instrumenting a Cessna T-41D with accelerometers, a GPS module, and a yoke-tracking system provided values close to those found with flight test, using arbitrary data instead of predefined test points. This process provides values at the correct order of magnitude and demonstrates this research at the proof-of-concept level, with estimates for short-period damping ratio with errors as low as 1.7%.