The paper proposes an unplanned island detection method in a microgrid with micro phasor measurement unit (μPMU). The μPMU extracts certain features, by using multi domain nature of discrete fractional Fourier transform, from measured voltage data and feeds it to a random forest classifier model to make decision of island or fault occurrence. The method was tested on an IEEE 34 node distribution network. The results show that the method is robust, reliable, cost effective and increases situational awareness. The proposed method has 99.8% accuracy, 99.89% security, 98.11% reliability, zero non detection zone and detection time of 49 ms.

本文提出了一种在具有微相量测量单元（μPMU）的微电网中进行非计划岛检测的方法。μPMU通过使用离散分数阶傅里叶变换的多域性质从测得的电压数据中提取某些特征，并将其输入到随机森林分类器模型中，从而确定孤岛或断层的发生。该方法已在IEEE 34节点分发网络上进行了测试。结果表明，该方法鲁棒，可靠，具有成本效益，并提高了态势感知能力。该方法具有99.8％的准确度，99.89％的安全性，98.11％的可靠性，零非检测区和49 ms的检测时间。