Modeling of Linear Switched Reluctance Motors Using Fuzzy Clustering Method

Due to the double salient structure of the stator and rotor, magnetic saturation and nonlinear characteristic of switched reluctance motor (SRM), it is difficult to extract a comprehensive and accurate analytical model of the motor for all rotor positions. On the other hand, the implementation of different control approaches and investigating the motor behavior under different loading conditions require an accurate motor model. In this paper, a new modeling method based on the adaptive-neural-fuzzy systems (ANFIS) using data clustering approach is proposed. Approximation of flux linkage characteristic of the SRM using traditional fuzzy approaches including Mamdani, Takagi-Sugeno-Kang (TSK), and ANFIS methods are presented at first and the advantages and disadvantages of each approach are discussed. Then, the fuzzy clustering method (FCM) based on the TSK fuzzy type is introduced and its performance, application and advantages in motor modeling are presented. Finally, using an ANFIS system associated with data clustering approach (as the fuzzy inference system), the flux linkage characteristic of the linear switched reluctance motor (LSRM) is extracted. By comparing the results obtained by finite element analysis (FEA) and the proposed method, the accuracy and effectiveness of the proposed method in motor modeling are evaluated.