Model Predictive Control with Modulated Optimal Vector for a Three-Phase Inverter with an LC Filter

• Published in: IEEE transactions on power electronics Vol. 33; no. 3; pp. 2690 - 2703
• Main Authors: Nguyen, Hoach The, Kim, Eun-Kyung, Kim, Ik-Pyo, Choi, Han Ho, Jung, Jin-Woo
• Format: Journal Article
• Language: English
• Published: New York IEEE 01.03.2018; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Digital signal processors; Disturbance observers; Dynamic response; Finite control set (FCS); Harmonic distortion; Inverters; Load modeling; Mathematical models; Microprocessors; model predictive control (MPC); modulated optimal vector (MOV); Modulation; Parameter robustness; Parameter uncertainty; Predictive control; Robustness; Signal processing; Space vector modulation; space vector modulation (SVM); Steady-state; Switches; three-phase inverter; Voltage control
• DOI: 10.1109/TPEL.2017.2694049
• ISSN: 0885-8993

This paper proposes an effective model predictive control (MPC) scheme using a modulated optimal vector (MOV) and finite control options for a three-phase inverter with an LC filter. Unlike other MPC methods, the proposed MPC strategy exploits the unconstrained optimal vector (OV) of the continuous-control-set (CCS) MPC to limit the control options for the unconstrained mode. First, the analytical OV is derived based on a least-squares optimization. If the input constraints are not violated, the OV is applied with a space vector modulation (SVM) technique like the CCS-MPC. Otherwise, the OV is scaled into the MOV and only three control options are online evaluated to reselect the control input. Experiments are conducted on a three-phase inverter test bed with a TI TMS320F28335 digital signal processor to validate the improvements of the proposed method, especially the robust performances and fast responses. The comparative results with the FCS-MPC show the superior performances of the proposed scheme with smaller steady-state error and lower total harmonic distortion due to the analytical OV with SVM, more robustness to parameter uncertainties due to the disturbance observer, and faster dynamic response due to the online reselection of control inputs.