A Parameter-Adaptive Predictive Control Strategy for Grid-Connected Inverters With Different Filter Topologies

This article presents a novel adaptive inverse model predictive control (IMPC) algorithm for grid-connected inverters that operates effectively across different filter topologies (L, LC, LCL, etc.) without requiring parameter knowledge. To make the controller applicable regardless of the filter's type, the dynamic model is first manipulated and written as a function of the common dynamic states in all filter types. The manipulated dynamic model is then written in black-box model form to overcome parametric uncertainty and unmodeled dynamics raised by changing the type of filter. Based on the available measurements (inverter current and grid voltage), the parameters of the model are estimated online and used to update the estimated dynamic model. The estimated dynamic model is then used by the IMPC to predict the optimal switching vector that optimizes a predefined cost function. The proposed controller aims to achieve robust performance with the inverter-side current and output voltage measurements only while maintaining a fast dynamic response under balanced, unbalanced, and distorted grid conditions. Experimental investigations are conducted to validate the proposed approach and prove its advantages, such as accurate active and reactive power tracking across different grid conditions for different filter designs.