Design and experimental validation of a Model Predictive Control strategy for a VSI with long prediction horizon

This work presents the design and experimental validation of a Model Predictive Control (MPC) strategy for a Voltage Source Inverter (VSI) used in a three-phase uninterruptible power supply (UPS) for critical loads. The solution is obtained in two steps: first an unconstrained MPC problem is solved, which computes the continuous values of the voltage that must be provided by the converter. Then, a PWM modulation scheme is used to define the switching sequence of the switches that generate the desired input voltage to the LC filter. The first stage implies a dynamic optimization while the second one is just a static relationship. The solution of the unconstrained MPC problem gives rise to an explicit control law that can be computed beforehand, allowing the prediction horizon to be easily extended. This can improve the performance of other MPC formulation that use very short prediction horizons. The effect of the length of the prediction horizon and the control-weighting factor over the system performance is also evaluated experimentally in the paper. This study addresses how these parameters should be tuned to minimize the error between the actual and the desired output voltage. The proposed control strategy has been tested on a UPS supplying a three-phase RL load. The experimental results show that the proposed MPC controller achieves high performance of the overall system, even in the case of a model mismatch.