Artificial Neural Network Control of A Standalone DC Microgrid

This paper proposes a novel artificial neural network (ANN) based control method, integrated with droop control, for control of an islanded DC microgrid. The ANN controller is trained based on ADP (approximate dynamic programming) using LM (Levenberg-Marquardt) algorithm. A FATT (Forward Accumulation Through Time) algorithm is applied to calculated Jacobian matrix. The ANN performance is evaluated by using switching models of power DC converters. Performance of ANN in DC microgrid shows that the proposed controller has the ability to maintain voltage stability of standalone DC microgrid and manage the power sharing among the parallel-connected distributed generation units. For different transient scenarios, the ANN controller in DC microgrids also performs very well to tolerate load disturbances and track voltage references rapidly.