Despite distinctive advantages such as fast dynamic, simple concept and ease of implementation, model predictive direct power control (MPDPC) suffers from some major drawbacks. First, in order to provide sinusoidal currents and low-ripple powers, very high sampling rates are required for MPDPC implementation. Furthermore, this method requires accurate knowledge of system parameters and its performance deteriorates in presence of parameter uncertainties. In this paper, an improved MPDPC is proposed, which features: 1) high current and power quality at low sampling frequencies, 2) less sensitivity to parameter uncertainties, particularly unknown grid inductance, and 3) voltage sensorless operation. The first and second features are achieved by incorporating the concept of switching duty cycles into the conventional MPDPC. In the proposed method, two voltage vectors are applied during a control period and their duty cycles are obtained by a fuzzy logic modulator. The proposed fuzzy logic modulator works based on the real and reactive power errors. Eventually, voltage sensorless operation is accomplished by designing a full-order observer. Thanks to voltage sensorless operation, the system volume and cost can be reduced. Through extensive hardware-in-the-loop tests, the superiority of the proposed MPDPC in comparison with two conventional MPDPCs is demonstrated.

Keywords