Multi-objective predictive control for three-phase three-level neutral-point clamped inverter

This paper describes the strategy for the implementation of model predictive control (MPC) to a three-phase multilevel inverter. Though MPC has been active since the 1960s, it has only gained traction in recent years owing to the increased processing capabilities of modern devices which now help to remove the obstacle of large delays in this scheme. Moreover, this method is far superior in control of multiple input multiple output systems, could handle constraints of operation, and removed the necessity of a modulator, which was required in the popular pulse width modulation technique (PWM). In this paper, to achieve this scheme, we developed a novel cost function to reduce distortions in current output along with switching frequency reduction and capacitor voltage balancing. This was achieved by the iterative nature of the MPC that provides the least error which was then implemented in the next step. The action of prediction for the MPC was further tuned by adjusting the sampling time. In the process of implementation of MPC, the mathematical model of the converter and load was developed and simulations were performed on MATLAB/Simulink to understand the effect of changing the weighting factor on the system response. In this manner, this paper demonstrated the degree of control over DC-link capacitor voltages and switching frequency. The merit of such a predictive control over conventional PWM-based proportional integral (PWM-PI) control was analyzed by comparing the distortions in their respective waveforms and dynamic responses. It is observed that our strategy is more effective in reference tracking and gives a faster dynamic response as compared to PWM-PI control. Also, MPC was implemented using Intel's DE0-Nano board, a field programmable gate array (FPGA) platform, and the results were analyzed to prove the feasibility of the strategy for field use.