Research On Model-Free Adaptive Control Of Electromagnetic Bearing-Rotor System

Electromagnetic bearing-rotor system is a complex nonlinear system which provides a non-contact support structure for rotor through electromagnetic bearing.It has the advantages of long life,low loss and low vibration.Because of its instability,strong coupling and difficult to establish accurate model,it is necessary to design a reasonable controller to realized stable suspension control of the AMB rotor system.Therefore,the controller of the electromagnetic bearing rotor system is studied as follows:(1)Because of the nonlinear characteristics of the AMB rotor system,a full format model free adaptive control(FFDL-MFAC)algorithm is introduced and the controller of magnetic levitation system is designed.FFDL-MFAC is a novel data-driven control method for a class of unknown nonlinear discrete-time systems since accurate physical model except only for the input/output(I/O)measurement data.(2)In order to realize the stability control of the system,two tuning methods for the parameters of FFDL-MFAC controller are proposed.Firstly,the pseudo gradient initial value of FFDL-MFAC controller is taken as the unknown tuning parameter.Because of the learning and adaptive ability of BP neural network,the initial weight value of BP neural network is determined by integrating the differential signal into BP neural network.The stability of the method is proved by Lyapunov criterion.Then,according to the similarity between the PD controller and FFDL-MFAC control algorithm,the step factor parameter value of FFDL-MFAC controller are determined.Finally,through the simulation and experiment on the magnetic levitation system,the FFDL-MFAC algorithm can realize the stable levitation control of the magnetic levitation system is proved.(3)The mathematical model of the electromagnetic bearing-rotor system is established when the magnetic pole coupling is ignored at the equilibrium point.Then,the simulation model is built by Matlab/Simulink software to simulate the electromagnetic bearing-rotor system with FFDL-MFAC controller and PD controller.The simulation results show that compared with the PD controller,the FFDL-MFAC controller reduces the time to achieve stable suspension of the electromagnetic bearing-rotor system by 0.12 s.(4)In order to reduce the influence of differential measurement error,a projection error compensation method is proposed.The differential measurement error of rotor displacement is reduced from 2.5 % to 0.16 %.Then the FFDL-MFAC controller suspension control is carried out on the electromagnetic bearing-rotor system experimental platform.The experimental results show that the rotor is realized stable suspension by the FFDL-MFAC controller and the vibration amplitude is less than 2 μm.