An Iterative Kernel-PCA Based LPV Control Approach for a 4-DOF Control Moment Gyroscope

Abstract

This paper presents an LPV output feedback control strategy for a 4-DOF control moment gyroscope (CMG). The LPV model constructed by linearizing the nonlinear model around moving operating points contains fifteen scheduling variables causing a massive computational complexity involved in controller synthesis. A practical iterative approach based on kernel principal component analysis (kernel-PCA) is proposed for reduction of scheduling variables and LPV controller synthesis. The LPV controller designed for a fixed rotor speed is applied in feedback to a full nonlinear model of the CMG at different rotor speeds. Simulation results show that the controller obtained through the proposed method ensures improved performance in terms of increased operating range for CMG with reduced control effort while maintaining the transient performance and tracking accuracy.