An effective and timely detection of islanding is a task of paramount importance to nullify the potential threat to field personnel and equipment damage. A passive islanding detection method based on modal current and adaptive boosting is proposed in the present work. The three-phase currents of the target distributed generation unit are converted into modal signals reducing the dataset. These modal currents are subsequently decomposed into mono-frequency components by employing empirical mode decomposition tool. The authentic mono-frequency components identified using correlation are transformed with the help of Hilbert transform. Various features like entropy, skewness, power, kurtosis, signal-to-noise ratio, and total harmonic distortion are obtained from Hilbert transform. Subsequently, these features act as the input for the adaptive boosting technique to categorize islanding and non-islanding classes. The results obtained explicitly demonstrate that the proposed methodology is highly accurate with reduced non-detection zone.

有效和及时地发现孤岛是消除对现场人员和设备损坏的潜在威胁的首要任务。提出了一种基于模态电流和自适应升压的无源孤岛检测方法。目标分布式发电单元的三相电流被转换为模态信号，从而减少了数据集。这些模态电流随后通过采用经验模态分解工具分解为单频分量。使用希尔伯特（Hilbert）变换对使用相关性确定的真实单频分量进行变换。从希尔伯特变换中可以获得熵，偏度，功率，峰度，信噪比和总谐波失真等各种特征。后来，这些功能充当自适应升压技术的输入，以对孤岛和非孤岛类别进行分类。获得的结果明确表明，所提出的方法在减少非检测区的情况下是高度准确的。