Hardware-in-the-Loop Simulation of a Multilevel B2B-NPC with Weightless MPC Driving an Induction Motor

On model predictive control, the weighting factors are often challenging to tune, which usually relies on trial-and-error methods. To bypass this inconvenience, this paper enhances a previously proposed weightless method applied to Finite Control Set Model Predictive Control (FCS-MPC). An induction motor driven by a neutral point clamped inverter on a Back-to-Back configuration, connected to the grid and allowing power regeneration is presented. On the control side, a weightless predictive current control is presented along with a novel strategy to build the torque reference based on the motor dynamics, which can also dismiss load torque measurement or estimation with a speed deviation lower than 0.1%. To ensure reliable results, the power circuit was assembled on a hardware-in-the-loop environment, whilst the control was embedded on a digital signal processor, bringing the simulation closer to a real power plant. Simulation results proved that the strategy could regulate the Direct Current (DC) link voltage and the rotor speed and that the machine still operates on all four quadrants with a power factor close to unity. The mean absolute percentage error presented during full load was 0.12% for the DC link voltage and 0.22% for the rotor speed. For the grid current, the Total Harmonic Distortion (THD) obtained is 1.00% during pre-charge of the DC link capacitors, whilst for the stator current, it was 0.76% during full load. The control strategy also proved to be more efficient than previous works and presented robustness under many parameters mismatching.