Tuning PI controllers for optimal migrogrid operation based on Artificial Neural Network

The Microgrid achieve nowadays a great importance in providing energy to isolated areas. This paper demonstrates a new application of the Artificial Neural Networks (ANN) for online modification of the inverter based microgrid operation. A vector cascaded control technique is utilized as an inverter control strategy, which depends on the Proportional plus Integral (PI) controller. The suggested ANN is used to adapt the controller constants instantly. The ANN is simulated using MATLAB program. The input training samples of the ANN is obtained by carrying out different working settings simulated by PSCAD/EMTDC program. The optimal output training data fed to ANN is obtained by finding the optimal PI constants utilizing Evaporation Rate Water Cycle Algorithm (ERWCA). The utilized objective optimization problem used by the ERWCA is created by the Response Surface Methodology (RSM). The working scenarios included in this paper are 1) system conversion from grid connected mode to stand alone one, 2) system exposure to three lines to earth fault in the stand alone mode and 3) The steady state operation at grid connected mode.