Evaluation of parameter estimation methods for unstable aircraft

Abstract

While continuing to be of wide interest, the estimation of aerodynamic derivatives from flight data of an unstable aircraft poses several difficulties. The objective of this article is to provide an overview of some of the more recently introduced estimation methods together with some of the conventional ones. Five options addressed here are 1) regression startup, 2) equation decoupling, 3) filter error, 4) output error with artificial stabilization, and 5) multiple shooting method. An evaluation is made based on the parameter estimates both from simulated and flight data. The various methods yield comparable estimates from simulated responses pertaining to the short period motion of an unstable aircraft. Their application to X-31A aircraft flight data in unstable flight regime brings out some practical aspects. It is found that the application of the conventionally used output error method with artificial stabilization and of the multiple shooting method requires considerable engineering effort, and can still pose difficulties when analyzing longer duration maneuvers. The estimates from the filter error and equation decoupling methods compare with each other well. It is demonstrated that they provide an attractive and less laborious alternative to analyze longer duration maneuvers, e.g., consecutive elevator doublets or sweep inputs up to 30 s.