Abstract:

In grid-connected power generation systems, dc-link voltage control is needed to prevent energy losses, reduce voltage fluctuations and provide a stable energy flow. In addition, control of the power factor via the voltage source inverter is a process that supports the efficient use of the energy produced. Moreover, keeping the total harmonic distortion (THD) of the current injected into the grid in accordance with IEEE-519 harmonic standards (<5%), will increase the quality of the grid electrical current. In this study, three different methods (cascade pi-based, adaptive neuro fuzzy and artificial neural networks methods) were proposed for dc-link control. In addition, due to its high power factor and low harmonic distortion performance, a three-level neutral point clamped (NPC) inverter is modeled for grid-tied proton-exchange membrane fuel cells (PEMFC). The rated power of the proposed system is 50 kW and the system was tested under five different operation scenarios. According to the performance results, the THD in the grid current has been reduced from 8.02% to 3.52% compared to traditional methods, dc-link voltage ripple was observed to be around 1V (<1%), and also the power factor regulation performance increased as unity (>0.99).

Keywords: