A minimum sensitivity design method for output feedback controllers

Abstract

Introduction I recent years considerable attention has been focused on developing methods for control of large space structures (LSS). Three major problems arise when designing controllers for LSS. First, many of the available controller design methods assume that the full-state vector (or at least the estimated full-state vector) is available. In most practical situations these control laws encounter difficulties when only a low-dimensioned output vector is available. The real-time implementation of high-dimensioned state estimators, although perhaps feasible, remain computationally unattractive. Second, it is difficult and very expensive to design controllers for high-order systems such as a flexible structure. Therefore, it is usually desirable to design a controller based on a reduced-order model, which includes only the most important subset of the vibration modes, although we may retain the high-order model for validation purposes. The reducedorder-based controller may destabilize higher-order modes due to truncation and spillover effects. Therefore, we must include consideration of truncation and spillover effects when designing controllers based on a reduced-order model for LSS. Third, there exists a modeling error for any mathematical model of an LSS. These unknown, unmodeled effects can potentially destabilize the closed-loop system. In order to tolerate model errors, we need to optimize