Fast islanding detection using intra harmonic analysis and support vector machine based root cause identification in grid connected DC microgrid

Abstract

This paper deals with the issues related to the protection of grid-connected DC microgrid by islanding it under faulty conditions. The magnitude of fault current of a renewable energy penetrated DC microgrid remains low as compared to the conventional grid. Accordingly the detection of faulty conditions of DC microgrid is itself a challenging task. The present work provides an approach where the faulty condition can be detected on nearly real time basis with the help of the harmonic analysis of the current at the point of common coupling. The integer and inter-harmonic of current signal upto 11th order are analysed and compared with the specified standard values in IEEC6216 for diagnosis purposes. The integer and inter-harmonic of current signal upto 11th order are analysed and compared with the specified standard values in IEEC6216 for diagnosis purposes. The DC microgrid can then be islanded once the result of this comparison exceeds the threshold value for protection purposes. Besides islanding, the root cause of the fault has also been identified by analyzing the fault using support vector machine classification technique. The proposed scheme is simulated and verified extensively with all possible faulty conditions. It is also verified with a hardware prototype. The results of these analyses justify the adopted scheme where 99.23% accuracy is achieved.