Computationally efficient multistep model predictive control in cascaded H-bridge multilevel converter fed induction motor

Abstract

 In model predictive control (MPC) strategies for power electronics applications, extending the prediction horizon length has yielded significant benefits. Previous studies have approached the adoption of multistep predictive control (Multistep MPC) for cascaded H-bridge inverters. This method is used very effectively with applications that the prediction horizon is determined. However, when the number of prediction
steps is large, the calculation time will increase significantly. That seems to be the biggest disadvantage when applying the MPC control method. Therefore, this paper proposes an alternative new algorithm that improves the computational time of the predictive control algorithm. This algorithm helps to find control signals faster by parallel search strategy instead of sequential one as in traditional methods. The simulation results demonstrate the calculation speed is significantly faster, the control quality of the two methods is the same, and especially work well for large number of prediction steps applications that the conventional algorithm could not perform.