Nonlinear Regression Methods for Estimation

Regression techniques are developed for batch estimation and applied to three specific areas, namely, ballistic trajectory launch point estimation, adaptive flight control, and radio-frequency target triangulation. Specifically, linear regression with an intercept is considered in detail. An augmentation formulation is developed. Extensions of theory are applied to nonlinear regression as well. The intercept parameter estimate within the linear regression is used to identify the e ects of trim change that are associated with the occurrence of a control surface failure. These estimates are used to adjust the inner loop control gains via a feed-forward command, hence providing an automatic recon gurable retrim of an aircraft. The regression algorithms are used to consider reduced information applications, such as initial position target determination from bearings-only measurement data. In total, this dissertation develops algorithms for batch processes that broaden the envelope of successful estimation within the three aforementioned application areas. Additionally, the developed batch algorithms do not adversely impact the estimation ability in cases that are already estimated successfully by conventional approaches.