Abstract--- Model predictive current control (MPCC) with optimal duty cycle has been widely studied and applied in the field of permanent magnet synchronous motor (PMSM) drive due to its advantages of simplicity, fast dynamic response and good steady state performance. Recent research on two-vector MPCC (TV-MPCC) has shown that the pulse arrangement of placing a nonzero vector on both sides of a zero vector results in smaller current harmonics and lower switching frequency than the traditional pulse arrangement that places a zero vector on both sides of a nonzero vector. However, the state-of-the-art method still fails to obtain the optimal steady-state performance, because the freedom of vector combinations and vector duration optimization are not fully utilized. This paper proposes an improved two-vector model-free predictive current control (TV-MFPCC) method. An adaptive ultralocal model is adopted to achieve strong robustness. The proposed TV-MFPCC takes the combination of two nonzero vectors into account, thereby improving the steady-state performance at high speeds. It further optimizes the time allocation of the voltage vector placed on both sides, obtaining the minimal current total harmonic distortion (THD). Compared to the state-of-the-art TV-MFPCC method that involves placing a nonzero vector on both sides of a zero vector, the proposed TV-MFPCC not only produces slightly lower switching frequency, but also reduces the current THD significantly in the entire speed range, which can be more than 50% at high speeds.

Permanent magnet synchronous motor(PMSM),predictive control,current control,vector duration optimization