2026/05/12 by Rongkui Mei, Dehong Zhou, Xin Liu +2
Engineering · #Multilevel Inverters and Converters #Advanced DC-DC Converters #Microgrid Control and Optimization
paper · doi:10.1109/tpel.2026.3691939
Wide-bandgap devices enable high-frequency power converters but pose significant computational challenges for digital control systems. Multi-rate model predictive control (MR-MPC) tackles this challenge by decoupling sampling and switching operations, but its multi-step prediction mechanism leads to prediction error propagation within sampling intervals. However, conventional disturbance observers can only perform disturbance rejection at sampling instants, failing to provide suppression for the multiple time-series control inputs within each sampling interval. To this end, a time-series disturbance observer-based MR-MPC (TDMR-MPC) is proposed to enhance robustness against parameter mismatches. The proposed scheme utilizes a generalized proportional-integral observer to estimate time-series disturbances and embeds them into the predictive model for real-time suppression at each control point. Additionally, Lagrangian interpolation is proposed to deal with the dimensional mismatch between reference trajectories and control inputs. The stability of the system is rigorously proven, and experimental results on a silicon-carbide-based motor drive system validate the effectiveness of the scheme.