Real-time analysis of islanding detection scheme developed for AC microgrid system

In this study, a communication assisted passive islanding detection technique is proposed for microgrid having converter based distributed generation (DG) resources. In the proposed approach, the matrix pencil method is applied to reconstruct the voltage signal then both the raw signal and reconstructed signals are used to estimate phasor magnitudes using Discrete Fourier Transform (DFT). Full cycle DFT is unable to eliminate the decaying DC component completely in one cycle from the actual input signal, whereas the reconstructed signal is a filtered signal. Hence, subsequent to the application of DFT, the difference between actual and reconstructed signal becomes negligible during steady state condition but exists only if an event occurs in the microgrid. Utilizing the well-known fact presence of DC component in initial stage at the time of transient event, the magnitude difference of raw and reconstructed signals is computed at a given end. Accordingly, an index for each phase is calculated considering difference from both sides; point of common coupling (PCC) and DG. The target DG gets detected as being islanded from the main grid, if the indices of all the three phases remains positive i.e., greater than zero. The results obtained for tested conditions indicate that all the three indices of phases retain simultaneously positive sign only for islanding event. On other hand, non-islanding cases involve at least one or more phases indices being of negative sign. The proposed approach is tested on signals generated from simulation model. Also, the obtained results for signals from real-time test bed executed using Opal-RT, proves the superiority, robustness, and reliability of the proposed islanding detection algorithm.