Use of data-mining for non-invasive signature recognition in micro-grids: A preliminary approach applied to residential areas with PV converters

The future of distribution networks tends more and more to include computational power, embedded intelligence and smart metering on the high voltage level as well as the low voltage micro-grids. Several hardware solutions were developed to implement the so-called smart grids with measurement devices delivering data about the state of networks on various levels. This work introduces the use of a specific electric signature based on harmonic response of power converters in order to be able to get information in a non-invasive manner. A simulated residential grid with several loads and PV converters has been run real-time with 1 μs sampled data, for being able to retrieve information through data mining methods.