LFT-LPV Modeling and Control of a Control Moment Gyroscope

Techniques for the systematic modeling and Linear Parameter-Varying (LPV) control of highly nonlinear plants are proposed and applied to a four-degree of freedom (4-DOF) Control Moment Gyroscope (CMG). First, the factorization problem to yield LPV representations from general nonlinear models is formalized and tractable heuristics are proposed. Based on these, LPV parameter sets are automatically derived via a Principle Component Analysis (PCA)-based method that only requires model coefficients and allows straightforward approximation. Improved State-Feedback (SF) controller synthesis conditions for descriptor Linear Fractional Transformation (LFT)-LPV models are proposed that significantly reduce the synthesis effort. Both approximate and exact controller syntheses yield high performance in the entire operating envelope and are validated by experiments.