Model-Free Predictive Control for Grid-Connected Converters With Flexibility in Power Regulation: A Solution for Unbalanced Conditions

This article proposes a model-free predictive direct power control (PDPC) for grid-connected converters under unbalanced grid conditions. A new form of voltage vector (VV) selection is developed in the method that directly uses the time derivatives (slopes) of the active and the reactive powers. The slopes are predicted free of model parameters, lookup tables, and recursive calculations. An extended state observer (ESO) is designed and utilized for sampling data estimation to tackle the noise problem. The proposed method does not face any data-lacking problems during the startup due to directly controlling the active and reactive power through a switching logic. It is proposed to use new forms of power references for obtaining sinusoidal grid currents and flexibility in power oscillations control without any current and voltage sequences extraction to meet the grid standard and code requirements. The method is evaluated through simulation and experimental tests, and its performance is compared with that of conventional model-based and improved model-free predictive control (MFPC) methods. The results show the effectiveness of the proposed method and its superiority over the other methods.