Magnetization state estimation in variable-flux PMSMs

Variable flux permanent magnet synchronous machine (VF-PMSM) traction drives are capable of significant loss reduction over an electrical vehicle drive cycle using dynamic magnetization state (MS) manipulation. A continuous, online MS estimation method is desirable for VF-PMSM drives to optimally utilize the loss minimization capability with precise MS control. This paper presents a MS estimation method that uses flux observers and a structured neural network (SNN) based machine model to achieve high accuracy with low sensitivity to load current and a fast dynamic response. The steady state performance of the proposed method in the presence of load current is evaluated experimentally in the current plane as well as in the torque-speed plane. The effect of magnet temperature on MS estimation accuracy is also evaluated experimentally. The transient performance of the proposed method during MS manipulation transients is evaluated using FE methods.