An Ultra-fast Method for Analyzing IPM Motors at Multiple Operating Points Using Surrogate Models

Abstract

Interior permanent magnet (IPM) motors are frequently used for electric vehicle traction applications because of their high nominal operating region efficiency, power density, and flux weakening capabilities. However, due to the inherent cross-coupling and saturation effects of the direct- (d-) and quadrature (q-) axis parameters, it is difficult to predict the peak-motoring operating envelope in the flux weakening region because it is difficult to determine the maximum torque per ampere (MTPA) and maximum torque per volt (MTPV) trajectory without expending significant computational resources. Concordantly, without the MTPA/MTPV trajectory, and without loss data, it is difficult to estimate the motor operating efficiency at any general operating point. This issue is especially significant when in the early design stages or during optimization of a traction motor. This manuscript introduces a surrogate modelling technique and training heuristic for mapping the d- and q- axis parameters as well as hysteresis and eddy current losses at all motor operating points in a manner more computationally efficient than previous methods while maintaining acceptable accuracy. Thus, this method enables ultra-fast prediction of the entire peak motoring operating envelope and efficiency map of the electric machine at all operating points–even in the field weakening and deep field weakening regions. The newly proposed method is benchmarked against previous methods by comparing the predicted motor performance against experimentally-calibrated data. It is observed that the new method can predict performance with minimal percent efficiency difference over a wide range of operating points using 75% less computational resources than previous approaches.