Controller Design for Magnetic Levitation System

Abstract

Magnetic Levitation is a method by which an object is suspended in air by means of magnetic force. Earnshaw stated that static arrangements of magnet cannot levitate a body. The exception comes in case of diamagnetic and superconducting materials and by controlling magnetic field by control method. Diamagnetic materials or superconducting materials when placed in magnetic field produce magnetic field in opposite direction. Here the problem of controlling the magnetic field by control method is taken up to levitate a metal hollow sphere. The control problem is to supply controlled current to coil such that the magnetic force on the levitated body and gravitational force acting on it are exactly equal. Thus the magnetic levitation system is inherently unstable without any control action. It is desirable to not only levitate the object but also at desired position or continuously track a desired path. Here a linear and two nonlinear controllers are designed for magnetic levitation system. First a robust adaptive backstepping controller is designed for the system and simulated. The simulation results shows tracking error less than 0.0001m. The immeasurable state present is estimated by Kreisselmeier filter. The Kreisselmeier filter is a nonlinear estimator as well as preserves the output feedback form. However the control output is too high. To counteract the above problem Nesic backstepping controller is designed for the system by taking Euler approximate model of the system. The controller output is well within the range of 0.5~1 voltage. The reference tracking is also verified in simulation and the tracking error comes in range of 0.00015m. A linear controller is also designed for MagLev system as the region of operation of magnetic levitation setup is too small. A two degree freedom (2DOF) PID controller is designed satisfying a desired characteristics equation. The controller parameters are obtained by pole placement technique. The 2DOF PID controller is simulated and experimentally validated and it is seen that better result are obtained in 2DOF PID than 1DOF PID controller.