the Large Angle Magnetic Suspension Test Facility (LAMSTF). The modeling, design, analysis,.pdf

ROBUST TRACKING CONTROL OF A MAGNETICALLY SUSPENDED RIGID BODY Kyong B. Lim and David E. Coxy NASA Langley Research Center Hampton, Virginia SUMMARY This study is an application of H1 and -synthesis for designing robust tracking controllers for the Large Angle Magnetic Suspension Test Facility (LAMSTF). The modeling, design, analysis, simulation and testing of a control law that guarantees tracking performance under external disturbances and model uncertainties is investigated. The type of uncertainties considered and the tracking performance metric used is discussed. This study demonstrates the tradeo between tracking performance at low frequencies and robustness at high frequencies. Two sets of controllers were designed and tested. The rst set emphasized performance over robustness while the second set traded o performance for robustness. Comparisons of simulation and test results are also included. Current simulation and experimental results indicate that reasonably good robust tracking performance can be attained for this system using multivariable robust control approach. 1 Introduction The Large Angle Magnetic Suspension Test Facility (LAMSTF) has been assembled by NASA Langley Research Center for inhouse research in magnetic suspension technology. Reference [1] and the references therein give a detail description of the facility and discuss in detail the open-loop dynamic properties of the magnetic suspension system. This system represents a scaled model of a planned Large-Gap Magnetic Suspension System (LGMSS). Robust tracking control for the LAMSTF consists of controlling the attitude and position of the suspended rigid body in the presence of external disturbances and model uncertainties. The motion of the suspended rigid body is in general nonlinear and hence the linear, time-invariant perturbed motion about an equilibrium state is considered in this study. The underlying assumption in this study is that suciently accurate nominal and uncertainty mo