Modeling of a bearingless permanent magnet synchronous motor using adaptive weighted least square support vector machine

Abstract

A bearingless motor (BM) is an electrical machine with a built-in magnetic bearing element. The stator of the machine is fitted with two sets of windings with different pole-pair numbers, and the pole-pair numbers must be consecutive. The principle of BMs can be applied to most of the conventional motor types, i.e., reluctance motors, induction motors, and permanent magnet synchronous motors (PMSMs), etc. Among these types of BMs, the bearingless PMSMs (BPMSMs) are receiving more and more extensive attentions recently due to the remarkable advantages including reliability, high power density and high efficiency [1]. Because of the nonlinearity of the magnetic core, the resultant flux linkage is a highly nonlinear function of the rotor angle, and torque winding and suspension winding currents. The model established by conventional analysis methods can not reflect its real characteristics, and the control precision and the operational performance of BPMSMs are influenced [2]. Therefore, this paper uses a novel method based on adaptive weighted least square support vector machine (AW-LSSVM) regression algorithm to model the nonlinear flux linkage for a BPMSM.