Assessment of typical residential customers load profiles by using clustering techniques

Nicola Natale;Fabrizio Pilo;Giuditta Pisano;Matteo Troncia;Fabio Bignucolo;Massimiliano Coppo;Nicola Pesavento;Roberto Turri

2017-01-01

---

Abstract

This paper proposes a methodology for finding typical load profiles for residential customers by using clustering techniques. Such task is particularly challenging due to the great diversity of electricity use by residential customers. Specific characteristics of this kind of customers, as number of inhabitants or house surface, may help the clustering, but such features are often, maybe always, unknowable. In the paper, geographic information of customers, always known, has been used as first discriminant for grouping homogeneous daily profiles. Then, different clustering algorithms have been applied to a database of real customers and compared to verify their effectiveness. Finally, to validate the proposed approach ad hoc validation tests has been performed by considering a new database not used for clustering.