Composite Dead-beat Predictive Current Control of PMSM

To address the effect of parameter variation on the dead-beat predictive current control, this paper proposes a composite dead-beat predictive current control strategy. Firstly, the traditional dead-beat predictive current control and the traditional longberg observer are introduced. Subsequently, the effect of motor nominal value mismatch on the predictive current control and longberg observer gain is analyzed, the results show that the motor parameter error affects the current in the d-q axis, and the excessive flux linkage error causes large current fluctuations. To simplify the complexity of calculating the gain of the full-order longberg observer, the reduced-order longberg observer is designed, and the computation of the gain is largely simplified. Meanwhile, to eliminate the effect of flux linkage on the q-axis current, a reduced-order incremental longberg observer is designed, and the effect of the flux linkage variation on the q-axis current is effectively solved. To address the effect of the nominal value error, the parameter identification is introduced and the composite dead-beat predictive current control strategy is proposed. The nominal value of the motor is identified by the recursive least squares identification, then input into the controller to correct the parameters of the controller. Finally, the simulation validation in Matlab / Simulink is performed, and the results show that the composite dead-beat predictive current control has a faster current response speed and smaller current fluctuation, and the effect of flux linkage variation on the motor system is eliminated.