Quickest detection of unknown power quality events for smart grids

In this work, we study a change-point approach to provide the quickest detection of power quality (PQ) event occurrence for smart grids. Despite that both the occurrence time and the PQ event type are unknown beforehand, knowledge of the statistics of post-PQ event signals is required to implement the change-point approach. To circumvent this obstacle, we propose to model the unknown PQ events using different statistical distributions, namely the Gaussian, Gamma and inverse Gamma distributions. It is shown by computer simulation that all distributions under consideration can provide accurate PQ event detection. In particular, the inverse Gamma distribution demonstrates the most promising performance in our simulation.