Decoupling control of bearingless permanent magnet-type synchronous motor using artificial neural networks-based inverse system method

Abstract

A bearingless permanent magnet-type synchronous motor (BPMSM) is a complicated non-linear and strongly coupled system. The decoupling control among the electromagnetic torque and radial suspension forces, and the dynamic decoupling control between the two orthogonal suspension forces are key techniques for the stable operation of a BPMSM. In this paper, a comprehensive system model, which consists of two Park inverse transformations, two Clark inverse transformations, two current following inverters and load model of BPMSM is given. The reversibility of the complex-controlled object is proved. Combining an artificial neural network (ANN) based inverse model, which consists of a static ANN and five integrators with the controlled object, the control system is decoupled into two independent second-order linear subsystems and a first-order linear subsystem, i.e., two displacement subsystems and a rotor speed subsystem. It then becomes much easy to design the close-loop linear regulators to control each of the subsystems. The simulation test results have shown that the proposed method can achieve strong robustness, good static and dynamic decoupling performance.