A Novel BVC-RBF Neural Network Based System Simulation Model for Switched Reluctance Motor

Developing a precise system simulation model is a critical step in the design and analysis of optimal control strategies for a switched reluctance motor (SRM). To achieve this objective, the following works have been done in this paper. 1) A 3-D FEA model based on double scalar magnetic potential method (DSMP) is developed for obtaining the distributions of SRM magnetic field, then the flux linkage characteristics are calculated by using enhanced incremental energy method (EIEM). 2) In order to enhance modeling accuracy of the nonlinear flux linkage, a new RBF neural network with boundary value constraints (BVC-RBF) is used for approximating, based on the calculated flux linkage data. 3) The nonlinear BVC-RBF based simulation model of the SRM system is established for dynamic analysis with the power system block (PSB) modules of Matlab/simulink. 4) Simulation and experimental results are presented and compared for model validation. The validation study indicates that the developed model is highly accurate.