Potentials for reducing the power requirement of magnetic suspension systems by implementing a linear quadratic gaussian controller

Magnetic bearings provide an energy-saving and resource-sparing possibility for supporting rotors running at high speed. This paper studies the potential of reducing the energy demand of magnetic levitation systems. Practical investigations show that this is obtainable by combining an optimal state estimator with an optimal state space controller. This configuration is known as Linear Quadratic Gaussian (LQG) regulator. In a sequential way, all the steps required for implementing this control strategy are depicted, starting with describing and modelling of the system which is a fully magnetically levitated blower in this case. Subsequently, the realization of the system state estimator is delineated. Both hints for carrying out the discretization of the continuous time model and a simple approach for choosing the variance matrices of the estimator are given. Special attention is paid to the design of the state feedback controller. The influence of the weighting matrix elements on the controller performance is investigated as well as an integrative extension of the regulator which ensures good reference response behaviour. Finally, measurements concerning the energy consumption of the levitation system are presented. The results show that both in static and dynamic case the implementation of the LQG regulator yields reduced control energy effort, compared to a system controlled by a conventional PID controller. It is noteworthy that, simultaneously to the enhancement of energy efficiency, implementing the LQG controller improves the dynamic properties of the levitation system, e.g. transient time and overshoot.