Research On Circulating Current Control Strategy For Parallel Connection Of Multi-inverters In AC Microgrid

The AC microgrid with many advantages is the mainstream structure of the microgrid.A large number of inverter-type power supplies in the AC microgrid are connected to the grid in parallel,forming a parallel structure of multiple inverters in the AC microgrid.The stable operation of the parallel inverter system requires reasonable power distribution and sharing load current among the inverters.In the actual parallel system,due to the different system parameters of the inverter units,the equivalent output impedances of the parallel units are different,which makes the system generate circulating current and the power cannot be reasonably distributed among the inverter units.For the problem of circulating current,the virtual impedance is used to control the circulating current,and for the time-varying characteristics of the line impedance,the adaptive virtual impedance is obtained through neural network identification to better achieve the suppression of circulating current.Based on the structure of a parallel system with multiple inverter power sources,a simulation model is established.According to the characteristics of the parallel system,droop control is adopted as the control strategy of the inverter to achieve a reasonable power distribution under the ideal condition that the system capacity is inversely proportional to the line impedance and the droop coefficient.Based on the parallel structure of two inverter power sources,the relationship between droop control strategy and power distribution,the definition of circulating current and the relationship between circulating current and power distribution are analyzed,and simulation verification is carried out.According to the definition of circulation and its causes,the virtual impedance compensation method is adopted to compensate the difference in line impedances of the parallel system,so that the compensated equivalent output impedances are the same to control the circulating current.For the limitation of the fixed value virtual impedance on circulating current suppression when the line impedance changes,the self-adaptive virtual impedance based on BP neural network is adopted to achieve the suppression of circulating current.Under the condition of weak grid,since the amplitude of the equivalent line impedance increases greatly,and changes in real time with the load,the network structure and the weights of each layer may change.After the BP neural network is trained,the network structure and the weights of each layer have been determined.The improvement of this limitation requires a more complex network structure that will affect real-time performance,and BP neural network as a static network is difficult to accurately identify dynamic systems.Diagonal Recurrent Neural Network(DRNN)is used for online training to modify the weights of the network,and the accuracy of the real-time identification of line impedance is improved under the condition that the complexity of the identification network is unchanged.The simulation results prove that the adaptive virtual impedance based on the DRNN neural network can effectively improve the suppression effect of the circulating current of the parallel system.