Observer-based feedback for the magnetic levitation system

Abstract

Control of active magnetic bearings is an important area of research. The laboratory magnetic levitation system can be interpreted as a model of a single axis of bearings and is a useful testbed for control algorithms. The mathematical model of this system is highly non-linear and requires careful analysis and identification. The system is observable from position measurements as long as the electromagnet is powered as shown during the research. Practically measurable signals are the position and the coil current. The velocity that is necessary for any stabilizing control usually is obtained by numerical differentiation of the position. A more sophisticated approach is to estimate the velocity with an observer. Efficient observer types for this system are high-gain and non-linear reduced observers. The velocity estimated by an observer can be effectively used instead of a derivative in PID control of the position. Such an approach substantially improves control quality and extends the range of system’s stable operation. Even greater improvement is introduced by the addition of the non-linear feedforward to the control structure. The best results, provided the model parameters are correctly identified, are obtained with a control system consisting of the PID controller, the high-gain observer and the non-linear feedforward.