Microgrid protection using iterative filtering

This paper proposes an intelligent, nonpilot protection strategy for inverter-dominated microgrids using iterative filtering-based empirical mode decomposition (IFEMD) and extreme learning machine. The instantaneous frequency envelope (IFE) and the instantaneous Teager energy envelope (ITEE) of the most informative intrinsic mode function (MIIMF) obtained through processing the local current signal by IFEMD is used for fault detection and fault phase identification. But the disadvantage of IFEMD, like any other variant of empirical mode decomposition (EMD), is that for useful decomposition, it requires at least three cycles of data samples. For the above reason, a mathematical morphology-based dynamic event detection scheme is devised to pinpoint the fault inception instants which allows extraction of half-cycle postfault current IFE and ITEE. A fewer number of inputs to the classifier make the classification task faster and less computationally complex. The proposed scheme is extensively validated for both arcing and nonarcing faults with wide variations in operating parameters for different topologies and modes of operations of a standard microgrid model.