Intelligent Islanding Detection Scheme for Microgrid Based on Deep Learning and Wavelet Transform

Abstract

Microgrids operation mode automatically changes to islanded-mode in case of major disturbances in the main grid. Detection of unintentional islanding events is a significant challenge for microgrid operators. This paper proposes an efficient scheme for islanding detection of the microgrid. The proposed scheme is based on distributed generations (DGs) parameters. DGs parameters have been used as input of the proposed scheme. Discrete wavelet transform (DWT) is used to extract the hidden features of DGs parameters. Hidden features are the input of a deep neural network (DNN). The proposed scheme has significant accuracy in islanding detection, and it could be applied to different microgrids. We used three software to detect islanding events. First, DIg SILENT Power Factory is used to simulate operation cases of the microgrid. Then we used MATLAB for applying DWT to the measured parameters of DGs. Finally, we used a DNN created by TensorFlow as a machine learning technique.