Experimental Comparison of FPGA-Implemented Model Predictive Voltage Control to Cascaded Proportional Resonant Control for a Three-Phase Four-Wire Three-Level Grid-Forming Inverter of 250 kVA

Modern microgrid systems require inverters capable of forming an acceptable grid voltage, during islanded operation, both supplying and absorbing power, through unbalanced load conditions and transients. In order to address this concern, the steady state and transient performance of two control methods are compared which are both capable to supply unbalanced three-phase loads. The first method is a cascaded proportional resonant (PR) control, which is a state-of-the-art controller for single-phase-capable voltage-source inverters but suffers from lower performance in transient conditions. The second method is a new field-programmable gate array (FPGA)-implemented finite control set model-based predictive control (FCS-MPC) which shows high performance for transients as well as black-start capability. Experimental results show the model-based predictive controller (MPC) to have better steady-state performance, with an average total-harmonic-distortion (THD) over the entire steady-state operation range of 1.4 % vs. 2.2 % for the PR controller, and a voltage regulation error of 0.71 % for the MPC vs. 1.1 % for the PR controller over the entire operating range. The MPC also shows advantages over the PR controller during transient response conditions, enabling settling times within 600µs vs. 100 ms of the PR controller.