Robust Control And Neural Network Control Of Magnetic Bearings

Active control methods are emphasized in this paper. In view of the coupling between the freedoms of the bodies moving in the space, the new control system of rotor is designed, using the rigid dynamics theory to formulate equations of the rigid rotor. And the system is made up of a robust controller and a neural network controller.(1) The robust controller is based on the theory of Hx control. Mixed sensitivity approach is used to ensure the system to retrain the exoteric low interference at low frequency, and to be robust and stable when the system exist uncertainty at high frequency. And the steps of selecting weights functions is concluded according to abnormal Hn problems.(2) The neural network is used to further suppress the system uncertainty and error resulted from exoteric interference to improve the control precision by its good learning ability. The added neural network can be took on as the offset to the system, so the output of the neural network is constrained as to ensure the whole robustness of the system according the requirement of robustness.The calculated examples show that, compared with other conventional control methods, the methods used in this paper can ensure the system to have stronger robustness according to the system uncertainty, suppress impact interference more obviously and quickly, and further weaken the vibration resulted from low interference so as to improve the control precision.