Decentralized Control of Bidirectional Converters in a Grid-connected DC Microgrid to Increase System Stability Using a Genetic-Neural Algorithm

Using bidirectional parallel converters improves system reliability as they can bypass faults. Therefore, the bidirectional parallel converters structure is adopted for the DC micr ogrid connected to the grid. There is a circulating current between the bidirectional parallel converters that limits the overall capacity of the system and can damage switching devices. To solve this problem, in this paper, the mechanism of generating circulating current on the AC side when several bidirectional parallel converters work together was taken into consideration with the effect of constant power loads by the droop strategy Pdc-vdc2-f to share DC power and network frequency adaptation. The PI coefficients of the bidirectional parallel converters controller were optimized by online genetic algorithm and neural networks and updated with system conditions. Both the stable and dynamic relationships between Pdc and vdc2 are linear and did not create constant power loads of unstable poles for the system. The results indicated that compared with conventional idc-vdc droop control, the proposed Pdc-vdc2-f strategy can increase dc power control and system stability, consequently increasing the dc voltage regulation dynamics.