A new passive islanding detection approach using wavelets and deep learning for grid-connected photovoltaic systems

The renewable-based distributed energy resources (R-DERs) are extensively used in the smart distribution systems. Despite the advantages of the R-DER, they also introduce some operational challenges such as the unintentional islanding, reverse power flow, and several protection issues. Due to the unpredictable faults, an island or multiple islands can be formed throughout the electrical power systems. The island may have R-DER that shall be tripped within 2 s of the islanding formation as per IEEE 1547 and hence the islanding must be detected. This paper introduces a passive islanding detection approach that uses deep learning combined with the continuous wavelet transform and hence no need for identifying the islanding features a priori as in the existing islanding detection approaches. The numerical examples demonstrating the effectiveness of the proposed approach are presented and the conclusion is drawn.

Automated summary

This paper introduces a passive islanding detection approach that uses deep learning combined with the continuous wavelet transform, which can effectively solve the islanding detection problem without needing to identify islanding features in advance.