Neural network based NARMA controller for a DG in islanding mode

A neural-network based nonlinear autoregressive moving average controller is proposed for a DG in its islanding mode. When the DG is disconnected from the utility grid, its grid connected dq-current controller is disabled and the proposed controller undertakes control task. In the proposed controller, dynamics of the system is modeled and identified by two MLP neural networks. Then the controller is designed based on the constructed model. The performance of the controller is evaluated by simulations in SIMULINK/MATLAB.