Modeling of Bearingless Switched Reluctance Motor Based On Artificial Neural Network

Accurately calculating and modeling of the flux linkage characteristics is a critical step in design and analysis of optimal control strategies for Bearingless Switched Reluctance Motor (BSRM). But due to the highly nonlinear characteristics of BSRM, it is very difficult to derive a comprehensive mathematical model to satisfy the overall characteristics of the machine. To overcome this problem, in this paper, an edge finite element method (EFE) based 3-D FEM and a new enhanced incremental energy method were utilized for calculating the flux linkage characteristics; using the calculated flux data, an adaptive neural fuzzy inference system (ANFIS) based flux model was developed. Simulation results are presented and compared with the analytical nonlinear modeling method, which verified that the proposed model can modeling the BSRM more accurately and adaptively.