Efficient Fixed-Switching Modulated Finite Control Set-Model Predictive Control Based on Artificial Neural Networks

The disadvantage of finite control set-model predictive (FCS-MPC) is that the switching frequency variable and relies on sampling time operating point. This paper describes how to implement a new algorithm achieve fixed-switching functionality for FCS-MPC. used approach combines FCS-MPC with SVPWM, resulting in calculation dwell times selection best two active vectors next sample interval. These have significant impact performance during transient steady-state conditions, their values are determined using various mathematical models. To solve problem lower harmonics distortion compared conventional modulated MPC (M2PC), an ANN-based trained network proposed calculate duty-cycle applied thus ANN receives cost functions zero vector from M2PC determines optimal each based proper tuning. In this way, three goals achieved, first goal explicitly obtains frequency, secondly, as simple M2PC. Finally, feature including objectives non-linearity still applicable. paper’s case study level voltage source inverter (2L-VSI) uninterruptible power supply (UPS) applications. results MATLAB/Simulink revealed ANN-M2PC has retained all features addition at while quality significantly enhanced.