Dealing with islanded microgrids (MGs), this paper aims at improving the secondary control process to restrict the fluctuations in both the voltage and frequency signals. With the aim of retrieving these parameters at the nominal values, an intelligent control scheme is devised to adjust the corresponding control parameters. To do so, an on-line self-optimizing control approach is embedded in the MG’s central controller. In the tuning process, evolutionary-based techniques such as genetic algorithms provide proper initial adjustment for the parameters. Subsequently, an artificial neural network (ANN) is triggered to provide accurate online modification of the control parameters. Specifically, the training capability of the ANN mechanism along with its extensibility feature avoids the dependency of the controller on the operating point conditions and accommodates different changes and uncertainty reflections. Detailed simulation studies are conducted to investigate the performance of the proposed approach, and the results are discussed in depth.
