Luenberger Observer-Based Model Predictive Control for Six-Phase PMSM Motor With Localization Error Compensation

The reference vector strategy has been widely used in the model predictive control (MPC) to reduce the computational burden. However, the reference vector derivation as well as the predictive model are sensitive to the machine parameters. This article presents a robust MPC with Luenberger observer to compensate the localization error of the reference vector for a six-phase permanent magnet synchronous motor (PMSM) drive. First, the two-step synthesis technique is employed to increase the tracking capability in the primary subspace and reduce the harmonic currents in the secondary subspace. Then, the influence of the control set design on the robustness is investigated, and it is confirmed that the more densely the voltage vectors locate, the easier the erroneous vector selection occurs. The Luenberger observer is adopted to compensate the localization error of the reference vector caused by the machine parameter mismatch. Subsequently, the cost function is defined in the form of the voltage vector error directly, thus greatly simplifying the control structure. In this way, the steady state performance as well as the parameter robustness are enhanced significantly. Experimentations are conducted to verify the validity of the proposed method.