Islanding Detection in Grid-Connected Distributed Photovoltaic Generation Using Artificial Neural Network

Photovoltaic (PV) systems are nowadays one of the most wide-spread renewable energy systems in the network or grid with one purpose to improve the reliability of the grid. However, PV systems in the network also contribute a negative impact as well; when the main grid fails to supply the load and there is a part of the load energized by the PV systems while being isolated. This case is defined as islanding. If this condition cannot be detected, the load bus will experience voltage disturbance and power quality problem. This paper presents an islanding detection using Artificial Neural Network method (ANN). ANN learning data are generated from simulations under three main scenarios: power match, overvoltage, and undervoltage, with varying power factor (cos phi). Voltage signal at PCC node in load bus is classified to identify if system is in islanding condition or not. The simulation results shows that the built ANN is capable to detect both islanding and non-islanding mode.