Long-horizon direct model predictive control for a series-connected modular rectifier

This paper presents a long-horizon direct model predictive control for a series-connected modular rectifier. The topology combines a diode rectifier and an active-front-end (AFE) converter to achieve a modular dc railway power supply. Two formulations of the optimization problem, i.e., power and current control, are investigated. The current control problem-solved with the sphere decoder for reduced computational effort-is compared with the power control problem-solved with exhaustive enumeration-in terms of current distortions and distribution of the harmonic spectrum. The latter have to meet strict grid standards, such as IEEE 519 and IEC 61000-2-4 standards. As shown, thanks to the long prediction horizon the total demand distortion of the converter current can be reduced, while keeping the device switching frequency low due to the medium voltage target.