Model-Free Predictive Current Control of PMSM Drives Based on Variable Sequence Space Vector Modulation Using an Ultra-Local Model

Traditional deadbeat predictive current control (DBPCC) based on space vector modulation (SVM) is widely used in the field of high-performance control of permanent magnet synchronous motor (PMSM) drives due to its simple concept, fast dynamic response, and fixed switching frequency. However, the performance of DBPCC depends heavily on the motor parameter accuracy, which severely deteriorates when the machine parameter mismatches happen in practice owing to temperature, saturation, and so on. Furthermore, the traditional DBPCC uses only one vector sequence in the full speed range, leading to a relatively high proportion of current harmonics at high modulation indexes. To solve the problems above, this article proposes a model-free predictive current control (MFPCC) method for PMSM drives. Different from the prior ultra-local model, both the nonphysical parameter and the unknown dynamic part of the system are updated online by using the voltage and current in the last two control periods, which makes the proposed method universal and adaptive while achieving strong robustness. Next, the reference voltage vector is calculated by the adaptive ultra-local model and synthesized by variable-sequence SVM (VS-SVM) to obtain minimal current harmonics. The proposed method is compared to the traditional DBPCC. The experimental findings support the idea that the proposed method offers better steady-state performance and stronger robustness and that it can lower the total harmonic distortion (THD) by more than 30% at high modulation indexes.