Cluster analysis for Power Quality monitoring

This paper proposes a technique based in the graphical cluster analysis to be implemented in Smart Power Quality Analyzers (SPQA), a proposal Instrument Class S, according to the standard UNE-EN 61000-4-30, with the goal of monitoring electrical grids where new events arise as a consequence of the penetration of new energy resources. In the presence of an electrical fault the instrument triggers the measurement procedure, according to a settled threshold of the PQ, and calculates the higher-order statistics (HOS) features in the time domain that enable the classification. Performance results over a battery of 160 signals show two clusters (sags and transients), with an acceptable accuracy of 80%. The work also presents the concept of the uncertainty inter-cluster area. The analysis of the intensity cluster disturbances helps the grid operator characterizing better the database and introducing some qualitative criteria for monitoring the smart grid.