Self-scheduled control of a gyroscope

Abstract

A classical mixed sensitivity minimization approach and a model matching formulation are compared with the goal to design a linear parameter-varying augmented state feedback control law for a laboratory-scale control moment gyroscope. Dynamic weighting filters are used to impose integral action and roll-off on the controller. Consequently, measurement noise is effectively suppressed and steady state accuracy is guaranteed even in the presence of input disturbances. Both designs are validated in real-time experiments and compared to a previous design that uses static weights. With the new designs, control effort is reduced while transient performance is maintained and tracking accuracy, as well as disturbance attenuation, is improved.