Name: GIOVANNA GOMES ANDRADE
Publication date: 10/09/2026
Examining board:
Name |
Role |
|---|---|
| DOMINGOS SAVIO LYRIO SIMONETTI | Examinador Interno |
| LUCAS FRIZERA ENCARNACAO | Presidente |
| RENNER SARTÓRIO CAMARGO | Examinador Externo |
Summary: This work presents a quantitative and qualitative comparison between two control strategies applied to the current control of grid-connected power converters: the traditional Finite Control Set Model Predictive Control (FCS-MPC) and a Sliding Mode-based Finite Control Set Model Predictive Control (FCS-MPC-SM). The comparison was conducted using Hardware-in-the-Loop (HIL) real-time simulations of a three-phase gridconnected inverter with an inductive filter. Performance was evaluated under both steady-state and transient operating conditions using settling time, total harmonic distortion (THD), mean absolute error (MAE), maximum error, and parametric sensitivity analyses. In addition, the influence of computational delay was investigated before and after the application of delay compensation. The results demonstrate that both control strategies achieve satisfactory current regulation. However, after delay compensation, the conventional FCS-MPC outperformed the FCS-MPC-SM under steady-state conditions, achieving a current THD of 3.33% compared with 4.54% for the FCS-MPC-SM, and a voltage THD of 79.66%, approximately 29.4% lower than the 108.70% obtained with the FCS-MPC-SM. Furthermore, the conventional FCS-MPC exhibited lower MAE and maximum error for most operating conditions, as well as lower sensitivity to parameter uncertainties, whereas the FCS-MPC-SM produced power MAE values up to three times higher during the parametric sensitivity analysis. On the other hand, the FCS-MPC-SM achieved a shorter settling time during transient operation. Overall, the results indicate that, although both strategies are suitable for the current control of grid-connected inverters, the conventional FCS-MPC provides greater robustness and superior overall performance when computational delay and parameter uncertainties are taken into account.
