Finite Control Set Model Predictive Control for Grid-Connected Parallel Power Converters With Guaranteed Optimality

Abstract

Model predictive control (MPC) has been applied to parallel power converters to achieve current tracking and zero-sequence circulating current (ZSCC) suppression. To facilitate real-time implementation, simplifications of the optimization problem are commonly made by reformulating the cost function or restricting the feasible control set. However, inappropriate simplifications can lead to suboptimal solutions, degrading control performance. To address this issue, this work formulates a formal finite control set MPC (FCS-MPC) problem for grid-connected parallel power converters. It obtains the optimal solution using the computationally efficient modified sphere decoding algorithm. The guaranteed optimality allows the proposed FCS-MPC to maximize the system control performance in suppressing ZSCC, reducing current distortion, and improving transient response. Furthermore, we propose an alternative cost function that includes the grid-side total current as one of the control variables. Owing to the interleaving-wise mechanism, there is a significant enhancement in the power quality of the grid integration current, particularly at lower switching frequencies. Finally, experimental results from a lab-constructed test bench validate the theoretical analysis.