Model-Free Predictive Current Control of a Multilevel Cascaded H-bridge inverter for Photovoltaic Systems

Model-based predictive current control (MBPCC) has been developed to control photovoltaic power plant current to allow high-performance accuracy and fast dynamic responses compared to the classical approach based on linear controllers. This type of controller also has many drawbacks and this paper deals with two of them: the need for a good prediction model and the need for the converter output voltage to have good closed-loop results. If MBPCC is not properly tuned, a variation of dc-link voltage or the use of a less accurate current prediction model reduces the ability of the controller to instantly track the current with high accuracy. To avoid such problems, model-free predictive current control (MFPCC) has been proposed only for two-level inverters up to date. However, it is known that it provides a reduced power quality compared to multilevel inverters for medium-voltage applications. To design an MFPCC to multilevel-VSI the challenge of stagnant conducting current which increases with respect to the number of levels in an inverter must be solved first. In this paper, a model-free predictive control based on extended adjacent states is proposed to control the grid current with a multilevel inverter. Through simulation results, a good tradeoff between current THD, CMV and switching frequency especially under parameter variations is obtained.