Islanding detection method for microgrid based on extracted features from differential transient rate of change of frequency

One of the most important challenges in microgrid operation is the unintentional islanding occurrence. Unintentional islanding can cause serious safety hazards and technical issues. Islanding detection methods can be classified into active and passive methods. The main disadvantages of the passive methods are large non-detection zone as well as determination of suitable threshold value to avoid unwanted distributed generations tripping in normal network events. In order to overcome these drawbacks, this study proposes a novel, fast, and reliable method to identify islanding conditions. The proposed method calculates different transient states in the rate of change of frequency signal in two consecutive cycles. Various features of the differential signal are extracted. The extracted feature vectors associated with different operation conditions such as islanding and non-islanding events are used to train artificial neural networks (ANNs). The performances of different structures of ANNs and also other machine learning methods such as support vector machine and adaptive neuro fuzzy inference system are evaluated for islanding detection purposes. The simulation results indicate that the proposed method provides more accurate and faster responses compared with other conventional islanding detection methods.