Decoupling Based On Inverse System And Digital Control For AC Hybrid Magnetic Bearings

Magnetic bearings have advantages of no abrasion,no lubricated and seal,no pollution,long life to the traditional bearings,so they are attracting more attention gradually.They also have widely potential application prospect in the area of machinery industry,energy transportation and aerospace.Supported by National Natural Science Foundation of China(50575099),our team is researching a new structure of electric spindle system,which is supported by AC hybrid magnetic bearing(HMB).Neural network inverse system method has advantage of good robustness and doesn’t depend on the object’s accurate mathematical model and parameters.So the neural network inverse system method is applied on the multivariate decoupling control of AC HMB. The mathematical model is set up and decoupling control algorithm is deduced.The detail contents are as follows:Firstly,on the base of introducing the configuration and working principle of an axial HMB, the suction equation of axial HMB system is set up.In order to improve the axial HMB mathematical model’s linearization precision and reduce the error from neglecting the above second order infinitesimal after Taylor expansion,α-th order inverse system method is used to make a linearization process on the mathematical model.Then the closed-loop controller is designed and simulation research is done.The simulation results have shown that accurate linearization control can be realized usingα-th order inverse system method for the axial HMB, and the whole control system has good dynamic and static performance.Secondly,on the base of introducing the configuration and the working principle of 5 degrees of freedom(DOF) AC HMB,the state equations of 5-DOF AC HMB are set up.According to the 5-DOF AC HMB’s characteristics of multivariable,nonlinear,strong coupling,neural network inverse system method is used to construct the inverse system of the 5-DOF AC HMB.Cascading the inverse system with the 5-DOF AC HMB system,the 5-DOF AC HMB system is decoupled into five independent linear displacement subsystems.Then linear system theory is applied to the pseudo linear system to synthesize and simulate.The simulation results have shown that neural network inverse system control strategy can realize dynamic decoupling control among 5 degrees of freedom of the system.Lastly,using digital signal processor(DSP) as CPU,the hardware of the digital controller has been designed.The flow charts for main program and every subprogram have been listed.The related software is developed and local experiment is done.The detailed experimental and testing procedures have been proposed,which establishes a good foundation for the next parameters debugging and function test.