Control of Shaft Vibrations on Magnetic Bearings Using Neural Network and Sliding Mode Controller

This article discusses a method for complete control of the dynamics of a rotating shaft on magnetic bearings under the effects of mass unbalance. The electromagnetic bearings used in this research are 2 four-pole bearing at the two ends of the rotor, which are actuated by differential currents. Full dynamic behavior of 3- dimensional rigid rotor and its effects on dynamic model are included. The effects of rotating mass unbalance are also included in the equations. The geometric couplings between electromagnetic forces of the coils are included as uncertainty. By using sliding mode controller and a neural network to estimate the system nonlinear-coupled equations, in a way suitable for sliding mode controller, a controller for this 4 input system is designed. Using a computer simulation of the real system, the difference between actual and estimated model and the effects of the controller are studied. For different operating modes the controller parameters are then determined. The present method allows the controller to apply necessary changes required by the time variant system equations.