Authors: D. Susitra; S. Paramasivam
Addresses: Faculty of Electrical and Electronics, Sathyabama University, Chennai 600119, India ' ESAB Group, Sriperumbudur Taluk, Kanchipuram District 602105, India
Abstract: This paper presents a non-linear inductance profile estimation of a Switched Reluctance Machine (SRM) using real-time applicable modelling techniques. The estimation techniques are based on regression analysis and Adaptive Neuro-Fuzzy Inference System (ANFIS). Mathematical models of phase inductance L(I,θ) using both the techniques have been successfully tested for various values of phase currents (Iph) and rotor positions (θ) of a non-linear SRM. It is observed that the proposed techniques are highly suitable for inductance L(I, θ) modelling of SRM, which is found to be in good agreement with the training data used for modelling.
Keywords: nonlinear inductance models; regression technique; ANFIS; adaptive neuro-fuzzy inference system; multiple regression; SRM; switched reluctance machines; power electronics; phase inductance profile; neural networks; fuzzy logic; mathematical modelling; phase currents; rotor positions.
International Journal of Power Electronics, 2014 Vol.6 No.3, pp.257 - 275
Received: 26 May 2013
Accepted: 10 Apr 2014
Published online: 02 Sep 2014 *