A self-training numerical method to calculate the magnetic characteristics for switched reluctance motor drives

Based on the two-dimensional (2-D) least squares method, this paper presents a novel numerical method to calculate the magnetic characteristics for switched reluctance motor drives. In this method, the 2-D orthogonal polynomials are used to model the magnetic characteristics. The coefficients in these polynomials are determined by the 2-D least squares method. These coefficients can be computed off line and can also be trained on line. The computed results agree well with the experimental results. In addition, the effect of the order number of the polynomials on the computation errors is discussed. The proposed method is very helpful in torque prediction, simulation studies and development of sensorless control of switched reluctance motor drives.