Unveil the Shape: Data Analytics for Extracting Knowledge from Smart Meters

摘要

The increasing availability of data gathered fromsmart meters makes it possible to develop specific data analyticsprocedures to extract knowledge from data. The analysis of hugeamounts of data coming from millions of users needs appropriateplatforms and efficient algorithms. The platform SHAPE(Statistical Hybrid Analysis for load ProfilE), set up by edistribuzione,has enabled to determine, for the first time in Italy,the consumption classes at national level based on the behaviourof low-voltage customers. This paper discusses the data handlingprocess in SHAPE, based on customer sampling by macrocategories,definition of consistent time periods, and clusteringbasedload profiling, and addresses specific applications such ascustomer classification, load prediction, probabilistic aggregationof residential loads, and sharing of the energy not supplied bymacro-categories of users.