Artificial Intelligence-Based Cyber–Physical Events Classification for Islanding Detection in Power Inverters

Along with the rapid integration of distributed generation units (DGUs) into the power grids is the rise in unconventional and unpredictable patterns of the undesirable cyber–physical intrusions and faults; this drastically increases the risk of islanding possibilities and threatens the sustainability of the energy delivery infrastructure. Classification of cyber–physical events and developing solutions to mitigate their impacts before rising to an islanding situation is a critical monitoring task in DGUs. Passive islanding detection has been widely applied to studying the behavior of voltage signals at the point of common coupling, which is a sophisticated challenge due to cross similarity among fault (event) patterns and their fast dynamics. In this article, a novel quadratic time–frequency decomposition, namely HSS-transform, is applied over an alternative complex representation of three-phase signal defined by the synchronous reference frame transformation. We further exploit the principles of informative sparse representation-based classification (TISC) to develop a comprehensive artificial intelligence framework for fast and reliable classification of DGU islanding and nonislanding events with the focus on practical limitations and requirements of a smart power electronics inviter as the desirable observational site. Different from the state-of-the-art techniques, TISC does not need any training procedure, while due to its linear mathematical formulation acts inherently fast with low computational burden on the inverter processing unit. Moreover, the simultaneous three-phase feature extraction strategy ensures preservation of the between-phase information.