Application of Multivariable Control System Design Methodologies to Robust Beam Control of a Space-Based Laser.

摘要

The complexity of large-scale dynamic systems provides a challenging proving ground for modern control system analysis and design techniques. High plant dimensionality inherent in large-scale systems can lead to breakdowns in numerics of state-space algorithms or intolerably long computational times, necessitating use of model reduction techniques. Reducing plant order consequently introduces unmodeled dynamics into the system, which must then be accounted for via stability and performance robustness considerations. A design framework is adopted herein which allows stability robustness to be guaranteed via unstructured uncertainty representation and the Small Gain Theorem, and performance robustness to be independently verified. The applicability of modern multivariable controller design techniques to large-scale systems is demonstrated by synthesis of robustly stable H2 optimal, H at infinity optimal, and H2/H at infinity loop-shaped compensators for a space-based laser forebody using reduced order models. Performance goals are expressed in terms of both allowable 2-norms and root-mean-square valves of line-of-sight and segment phasing errors, and an evolutionary process leads to a final controller design. First, ideally performing but non-robust unconstrained bandwidth H2 and H at infinity designs are presented which illustrate the danger of ignoring unmodeled high-frequency dynamics. Then robustly stable but poorly performing reduced bandwidth H2 and H at infinity designs are derived, achieved via constant penalizing of the system control at all frequencies.