Model predictive duty cycle control for three‐phase Vienna rectifiers

Abstract

The Vienna rectifiers are widely used in industrial applications. The computationally efficient optimal switching sequence model predictive control directly optimizes the switching sequences. Thus, it requires to store the switching states of 25 voltage vectors, which increases the memory usage. Moreover, the neutral-point (NP) voltage is regulated by redundant vector preselection. As only one redundant vector is applied during the whole sampling period, the NP voltage ripple is relatively large, and the grid current performance is still not satisfactory. To overcome these drawbacks, this paper proposes a model predictive duty cycle control strategy for three-phase Vienna rectifiers, which directly optimizes three-phase duty cycles. The three-phase duty cycles are first obtained by predicting the effects of three-phase duty cycles on instantaneous current variations and minimizing a cost function. To guarantee the grid current performance, a novel three-phase duty cycles modification method is proposed by considering the structural characteristics of the Vienna rectifier. To suppress the NP voltage ripple, the obtained three-phase duty cycles are further optimized. With the proposed method, low memory usage, low harmonic distortion in the grid current, and low NP voltage ripple are achieved simultaneously. Experiments are conducted to verify the effectiveness of the proposed method.

Sliding mode based adaptive linear neuron proportional resonant control of Vienna rectifier for performance improvement of electric vehicle charging system

“…where i o is mid-point current. Considering balanced DClink voltage, the DC-link capacitances, C 1 and C 2 are supposed to be equal, i.e., C 1 = C 2 = C. The mid-point voltage v mo of the VR can be given based on the upper and lower DC capacitor voltages [23]. The dynamic equation of the mid-point voltage is given as follows:…”

“…As such, the overall control solution is very computationally complex. In [23], a phase-locked loop (PLL) based model predictive control method has been applied to the VR. Despite possessing some excellent characteristics, this controller is not designed for an unbalanced grid operation.…”