Adaptive clustering method for LV customer profiling

The evolution of power systems, especially at distribution level, is particularly fast due to many novelties, as the impressive growth of distributed energy resources based on renewables and the economic drivers towards “full-electric” energy consuming scenarios (i.e. electric vehicles, electric air conditioning, cooking appliances, etc.). There is a general consensus in Literature that time series of consumption and generation are necessary to properly represent the impact of demand coincidence and of generation-load homotheticity. Thus, accurate modeling of production and demand becomes crucial to make rational hypotheses for network development and management, both at MV and LV level. Worldwide TSOs and DSOs use typical daily profiles for representing the consumption of the end-users. Updating these typical load profiles to the new behavior of current customers is often disregarded, and the used profiles refer to out-of-date and limited measurement campaigns. This paper proposes a clustering method for finding updated typical load profiles for different category of customers (household, commercial, industrial), able to be adapted if any change in the electrical behavior of customers occurs. In the paper results obtained by applying the proposed methodology to a large database of real consumption profiles of end-users in Italy have been discussed.