In this paper, a time-domain passive islanding detection technique for a grid-connected hybrid distributed generating system has been developed using modified empirical mode decomposition (MOEMD). With the help of the MOEMD, the point of common coupling voltage is decomposed into several oscillatory components called intrinsic mode functions. The islanding will be detected if the magnitude of the islanding detection waveform exceeds the threshold over a predefined delay time. The proposed technique can identify all kinds of islanding cases even for zero power mismatches nearly at 0.3 s of the predefined delay time. The effectiveness of the proposed scheme is tested for different operating conditions that include various non-islanding events. Moreover, the proposed method is faster than the discrete wavelet transform technique, and it has low computational time. The advantages of the proposed method are independent of the system parameters, zero non-detection zone, no classifier requirement, and its performance is independent of the capacity and nature of D.G. connected to the grid.

本文采用改进的经验模态分解（MOEMD）技术开发了一种用于并网混合分布式发电系统的时域无源孤岛检测技术。在MOEMD的帮助下，公共耦合电压的点被分解成几个称为固有模式函数的振荡分量。如果孤岛检测波形的幅度在预定义的延迟时间内超过阈值，则将检测到孤岛。所提出的技术甚至可以在接近预定义延迟时间0.3 s的零功率不匹配情况下识别出各种孤岛情况。针对包括各种非孤岛事件的不同操作条件，测试了所提出方案的有效性。此外，所提出的方法比离散小波变换技术要快，而且它的计算时间很短。该方法的优点与系统参数无关，零检测区为零，无分类器要求，其性能与与电网相连的DG的容量和性质无关。