Numerical Optimization of Magnetization State Manipulation for Variable Flux Memory Machine Under Voltage Constraint

The efficiency improvement of variable flux memory machines (VFMM) primarily depends on the adjustment of magnetization state (MS), which is realized by injecting a d -axis current ( i(d)) pulse with a suitable peak value. The reachable peak i(d) value is, however, severely restricted by the fixed dc bus voltage, especially at medium and high speeds. This article proposes a numerical optimization method based on the genetic algorithm (GA) to generate an optimal current trajectory for VFMM to conduct the magnetization manipulations at higher speeds before reaching the voltage capability of the inverter. First, a numerical optimization model of the dynamic MS manipulation process is newly established and then discretized by using the fourth-order Runge-Kutta (RK4) method, so that the current trajectory solving can be transformed to an optimization problem. Second, some constraints in this problem are identified and modified to simplify the solving process. Finally, the GA is used to obtain the preliminary optimized results under different parameters, based on which the optimal current trajectory is decided. Besides, the current controller is particularly modified to guarantee good performance under voltage constraint conditions. Experimental measurements on a hybrid-magnetic-circuit VFMM (HMC-VFMM) prototype verify the effectiveness of the proposed method.