An Inductance Online Identification Method for Model Predictive Control of V2G Inverter With Enhanced Robustness to Grid Frequency Deviation

Abstract

In this article, an inductance online identification method is proposed for model predictive control (MPC) of vehicle-to-grid (V2G) inverter with an enhanced robustness against grid frequency deviation. First, a full-order sliding mode observer is established to observe the virtual flux of the power grid. Second, a new inductance online identification method based on the dot product of the grid voltage vector and the observed virtual-flux vector is creatively proposed. A detailed theoretical analysis is carried out, which shows that the robustness of the proposed inductance identification method against frequency deviation can be enhanced by properly selecting sliding mode gains. Finally, by substituting the identified inductance into the MPC algorithm, the current control accuracy of the V2G inverter is improved accordingly. Detailed comparative experimental results verify the effectiveness and feasibility of the proposed strategy. When there is a frequency deviation of 5 Hz, which may occur in a weak grid system, the inductance identification errors of the conventional gradient correction method, the forgetting-factor least-squares method, and the direct calculation method are 2, 2.3, and 3 mH, respectively, while it is reduced to 1.75 mH for the proposed method.