This paper proposes a novel hybrid-magnetic-circuit variable flux memory machine (HMC-VFMM) by combining series and parallel hybrid magnet structures. Thus, the synergies of wide flux regulation range in parallel type and excellent on-load demagnetization withstand capability in series type can be simultaneously obtained with the proposed HMC design. Meanwhile, two sets of the permanent magnets (PMs) with high coercive force (HCF) and low coercive force (LCF), i.e. NdFeB and AlNiCo PMs, are employed to achieve high torque density and energy-efficient magnetization state (MS) adjustment. The topologies and tradeoffs of traditional parallel and series VFMMs are addressed first. In addition, the structure evolution, features and operating principle of the proposed HMC-VFMM are described, respectively. A simplified equivalent magnetic circuit is modelled to reveal the performance improvement of the machine. Then, the design improvements with q-axis barriers are presented to elevate the LCF PM working point for preventing the on-load demagnetizing effect, while maintaining the torque capability. The electromagnetic characteristics of the HMC design are investigated and compared with the parallel/series counterparts. Finally, the experiments have been carried out to validate the finite element (FE) analyses.

Model predictive control (MPC) has attracted a lot of attention for power converters and motor drives in the last decade, owing to its merits such as explicit concept, excellent dynamic performance, and decent applicability in digital implementation. However, as a model-based method, classical MPC is heavily dependent on the system parameters of the plant under control. In this article, a novel finite-set sliding-mode model-free predictive current control (FS-SM-MFPCC) is proposed for permanent magnet synchronous machine (PMSM) drives, which combines the theory of sliding-mode control with MPC. The proposed strategy is thoroughly model-free, irrelevant for the dc-link voltage, and computationally light. The original version of the suggested strategy suffers from two flaws: 1) poor steady-state performance and 2) existent tracking errors, which are solved by using a simple correction algorithm and an extended control set coordinating with an improved cost function, respectively. The comparatively experimental results based on a 500-W PMSM drive are given to confirm the benefits of the proposed strategy.

This article focuses on the position sensorless control for variable flux memory machine (VFMM) of which the magnetization state (MS) can be flexibly changed by applying a d -axis current pulse to achieve the so-called MS manipulation. First, the unique challenges in VFMM sensorless control are in depth analyzed and summarized based on the experimental results. An improved adaptive sliding-mode observer (SMO) and an improved phase-locked loop (PLL) are then proposed. Specifically, a newly developed dq -axis voltage pulse injection method is employed to accurately measure the dq -axis inductances of VFMM during MS manipulations and update them into SMO to increase its convergence speed. An adaptive law is proposed for tuning sliding-mode gains to obtain satisfying estimation performance of SMO under different conditions of speed and MS. In addition, to enhance the position tracking performance against the position estimation error resulted from MS manipulation or speed change, an improved PLL is structured, whose bandwidth can be adaptively adjusted by using fuzzy logic control. Finally, the feasibility and effectiveness of the proposed sensorless control strategy are verified through experimental measurements on a hybrid magnetic circuit VFMM prototype.