Artificial Neural Network PI-Based Compensator for DC-Link Current Control in Grid-Connected Photovoltaic Current Source Inverter

Conventional proportional integral (PI) controllers are the most, commonly, known and used techniques in grid-connected photovoltaic (PV) applications. Indeed, PI-based controllers are, practically, easy to design and implement. Moreover, they provide good performance, i.e., high-quality output signals, in steady-state conditions. Nevertheless, these controllers are, mainly, designed for the control of linear systems. Therefore, when implemented with non-linear systems, they suffer from slow dynamic response. Accordingly, in this paper, a non-linear and advanced PI-based controller is proposed. The proposed controller consists of an auto-tuned PI-based controller using artificial neural network (ANN). The main objective is to enhance the dynamic response of the DClink inductor current in grid-connected PV current source inverter (CSI) and to ensure that the injected grid current is equal to that generated by the DC-bus, which implies that the entire power provided by PV generator is injected into the grid. A numerical simulation model of the grid-connected PV system, i.e., CSI, PI-controller, and MPPT algorithm, is realized using Simulink and PLECS environments to validate the accuracy of the proposed solutions. The numerical results prove that the ANN auto-tuned PI-based controller leads to superior dynamic response of the DC-link inductor current and better-quality of the injected grid current compared to the conventional PI-based controller.