A Constant Switching Frequency Multiple Vector Based Model Predictive Current Control of Five-phase PMSM With Non-sinusoidal Back-EMF

Most of existing finite-control-set model predictive current control (FCS-MPCC) schemes for multiphase motor suffer from low order harmonic currents. Some virtual voltage vector ( $V^3$ ) based FCS-MPCC schemes can effectively suppress harmonic current by zeroing the harmonic subspace voltage on average during one sampling period. However, it fails when the motor has a non-sinusoidal back EMF. This paper proposes a constant switching frequency multiple-vector-based FCS-MPCC scheme. Unlike traditional FCS-MPCC schemes, the proposed scheme selects optimal $V^3$ s and their duty ratios in two orthogonal subspaces. Thus, it can simultaneously track the references in both orthogonal subspaces. In this approach, the optimal $V^3$ s and their duty ratios are directly obtained from the principle of deadbeat current control without complex and time-consuming enumeration-based state predictions and cost function calculations. In addition, the obtained optimal $V^3$ s and their duty ratios are adopted to rearrange the pulse sequence to obtain constant switching frequency and can be simply synthesized by carrier-based PWM. Furthermore, a discrete time disturbance observer (DTDO) is designed to improve the robustness of the proposed FCS-MPCC against parameter mismatch. Finally, comparative experiments with traditional MPCC schemes for five-phase permanent magnet synchronous machine (PMSM) are carried out to verify the effectiveness of the proposed scheme.