Robustness Analysis of Moving-Bank Multiple Model Adaptive Estimation and Control of a Large Flexible Space Structure

摘要

For this thesis, the truth model consists of 24 states, and the filter model consists of 6 states. The object is to determine whether or not the state mismatch between the filter and truth models seriously degrades or totally confounds the adaptation process. The system model used is a rotating two-bay flexible space structure, approximating a space structure that has a hub with extending appendages. The mass of the hub is greater than the mass of the appendage, and finite element analysis is used to develop the mathematical model. The system is developed in physical coordinates, transformed to modal coordinates, and the method of singular perturbations is used to provide a reduced filter model. The two dimensional parameter space consists of variations of the mass and stiffness matrices for the two-bay truss. Results indicate stable and accurate state estimation when the bank is initially centered on the true parameter. Stable control and accurate parameter estimation are not achieved. When the bank is initially centered at an incorrect parameter, stable and accurate state estimation is not demonstrated. These results indicate a total confounding of the adaptation process due to the unmodeled states, since accurate estimation and control can be generated when there are no such unmodeled states. Keywords: Kalman filters; Moving bank algorithms; Adaptive estimation; Multiple model adaptive control; Adaptive filters, Theses. (EDC)