Computation-Efficient Model Predictive Control With Common-Mode Voltage Elimination for Five-Level ANPC Converters

Abstract:

In this article, a computation-efficient model predictive control (MPC) is proposed to eliminate common-mode voltages (CMVs) of three-phase five-level active neutral-point-clamped (3P-5L-ANPC) converters. Originated from the CMV analysis of the 3P-5L-ANPC with all possible 125 possible voltage vectors, only 19 voltage vectors that generate zero CMV are adopted as the candidate voltage vectors for the MPC. The best voltage vector from the candidate voltage vectors is selected to track the current references. Then, appropriate switching combinations of the selected best voltage vector are determined to effectively balance the flying and dc-link capacitor voltages without any additional hardware components. Furthermore, the proposed MPC only chooses five candidate voltage vectors involving in MPC optimization according to the location of the reference voltage vector, which significantly alleviates the computational burden. Finally, the effectiveness of the proposed MPC in terms of the steady-state and dynamic performances is validated by simulated and experimental results.

Yong Yang;Huiqing Wen;Mingdi Fan;Menxi Xie;Simin Peng;Margarita Norambuena;Jose Rodriguez, “Computation-Efficient Model Predictive Control With Common-Mode Voltage Elimination for Five-Level ANPC Converters,” in IEEE Transactions on Transportation Electrification, vol. 6, no. 3, pp. 970-984, Sept. 2020. doi: 10.1109/TTE.2020.2996608
URL: http://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=9098961&isnumber=9200620