Autoregressive Moving Average Model-Free Predictive Current Control for PMSM Drives

To eliminate the influence of the parameter mismatches and obtain high model quality, a model-free predictive current control (MF-PCC) strategy based on the autoregressive moving average (ARMA) structure is proposed in this article and applied to the permanent magnet synchronous motor (PMSM) speed control system. Since the ARMA model group, which is a family of mathematical models containing AR, MA, and ARMA structures, considers operating states within several sampling periods to achieve better model accuracy, the plant is online-designed as this type, and its coefficients are estimated according to the sampled data by the normalized least-mean-square (NLMS) algorithm with adaptive normalized step length to achieve improved model quality with reduced calculation burden. Compared with the ultralocal MF-PCC strategy, the advantages of better stator current quality and robustness are demonstrated by the experimental results, as well as the reduced calculation burden compared with the recursive least square (RLS) algorithm used to estimate the coefficients.

Automated summary

In this article, a model-free predictive current control (MF-PCC) strategy based on the autoregressive moving average (ARMA) structure is proposed and applied to the permanent magnet synchronous motor (PMSM) speed control system to eliminate the influence of parameter mismatches and achieve high model quality. The ARMA model group, consisting of AR, MA, and ARMA structures, is considered to improve model accuracy by accounting for operating states within several sampling periods. The online-designed plant with estimated coefficients using the normalized least-mean-square (NLMS) algorithm achieves improved model quality with reduced calculation burden compared to the recursive least square (RLS) algorithm.