Magnetization State Control of Hybrid Magnet Memory Machines

Compared with conventional permanent magnet synchronous machines (PMSMs), the variable magnetization state (MS) of variable flux memory machines (VFMMs) offers improved machine efficiency and more flexible control. As a core part of VFMM control, the benefits of MS control have significant potential for promoting the application of VFMM drive systems. In this thesis, several novel MS control methods, and solutions to improve the performance of VFMMs are proposed and validated on a hybrid magnet memory machine (HMMM) control system.
Firstly, since a VFMM under different MSs has different PM flux linkages and inductances, a set of torque-speed curves can be obtained. The torques are the same on the interactions of these torque-speed curves, and MS manipulation on these intersections can be smooth and steady. An optimal demagnetization strategy is proposed to determine the best instants of demagnetization from the intersections of these grids. This solution is also verified during the remagnetization process. Secondly, to mitigate the speed fluctuation caused by the injected large magnetizing current pulse during MS manipulation, a novel dual magnetizing current controller is proposed. A q-axis current pulse is compensated to reduce the torque variation caused by the injected d-axis magnetizing current pulse. Thirdly, the circumcircle of the voltage vector hexagon is utilized as the modified voltage limitation and set as the boundary for MS manipulation to achieve the target MS, and a new MS control method is proposed. A unique MS control method utilizing the amplitude of demagnetization d-axis current pulse is therefore proposed to eliminate the potential unintentional demagnetization (UD) issue. When the amplitude of the d-axis current is bigger than the amplitude of the previous demand magnetizing current pulse, the MS manipulation will be processed. Thus, the UD issue can be avoided in the whole speed range. Finally, to achieve the MS close-loop control, a novel MS control method utilizing torque deviation is proposed. A model-based torque calculation is utilized for the torque estimation. When the torque deviation between the present MS and target MS meets the preset conditions, the type and the switch signal of MS manipulation will be generated.
The feasibility of all the proposed MS control methods are experimentally validated on a HMMM control system. The transient performance improvements of the proposed methods are significant, especially the proposed dual magnetizing current controller, which can reduce the speed fluctuation by more than 80% compared to conventional methods.