The application of support vector machines on decoupling control for bearingless synchronous reluctance motor

The bearingless synchronous reluctance motor has cross couplings among radial forces and electromagnetic torque, and between the two radial forces in x- and y-axis, and it is a complicated multivariate, nonlinear, and strong coupled system. Realizing decoupling control among radial forces and electromagnetic torque is a necessary condition of stable operation and precise control. In this paper, by using the feature of least square support vector machines(LS-SVM) in small sample learning and ability of universal approximation and identification fitting to get inverse model of bearingless synchronous reluctance motor, then the gained LS-SVM inverse models are in series before the original system to decouple a complex nonlinear multivariable system into 3 single input single output (SISO) pseudo-linear sub-systems. According to the linear control theory, the closed loop controllers are designed. The simulation results show that the system was decoupled, the overall hybrid control method has good dynamic and static performance, and the feasibility of the proposed control method is verified.