Estimation of SynRM Flux Saturation Model at Standstill using Artificial Neural Network

This paper proposes a method for estimating the magnetic flux saturation model of SynRM in a stationary state using an Artificial Neural Network (ANN). In the stationary state, the ANN is trained using the sampled current and the calculated magnetic flux obtained during hysteresis current control. The d-q axis magnetic flux generated according to the d-q axis current of SynRM appears symmetrically with respect to the axis and the origin. Using this phenomenon, the model was trained in the first quadrant by taking absolute values from the current and magnetic flux data. It was confirmed that the trained ANN model can represent the magnetic flux saturation phenomenon by comparing the estimated magnetic flux of the ANN model with the current-flux data. To verify the effectiveness of the proposed methods, the ANN flux saturation model was applied to sensorless drives with 1.5kW SynRM.