This study illustrates and discusses an original approach to classify the electricity consumers according to their daily load patterns. This approach exploits the notion of entropy introduced by Renyi for setting up specific clustering procedures. The proposed procedures differ with respect to typical methods adopted for electricity consumer classification, based on the Euclidean distance notion. The algorithms tested include firstly a classical method based on the between-cluster entropy and its slight variation. Then, a novel procedure is presented, based on the calculation of the similarity between centroids, with successive refinement to allow effective identification of the outliers. The outcomes of the classification carried out by using the proposed procedure are compared to the results of other available techniques, using a set of clustering validity indicators for ranking the clustering methods. On the basis of these results, it emerges that the novel procedure exhibits better clustering performance with respect to both the literature approaches and the classical entropy-based method, for different numbers of clusters. The results obtained are of key relevance for assisting the electricity suppliers in identifying a reduced number of load pattern dependent classes, to be associated with distinct consumer groups for load aggregation or tariff purposes.

Renyi entropy-based classification of daily electrical load patterns / Chicco, Gianfranco; SUMAILI AKILIMALI, Jean. - In: IET GENERATION, TRANSMISSION & DISTRIBUTION. - ISSN 1751-8687. - STAMPA. - 4:6(2010), pp. 736-745. [10.1049/iet-gtd.2009.0161]

Renyi entropy-based classification of daily electrical load patterns

CHICCO, GIANFRANCO;SUMAILI AKILIMALI, JEAN
2010

Abstract

This study illustrates and discusses an original approach to classify the electricity consumers according to their daily load patterns. This approach exploits the notion of entropy introduced by Renyi for setting up specific clustering procedures. The proposed procedures differ with respect to typical methods adopted for electricity consumer classification, based on the Euclidean distance notion. The algorithms tested include firstly a classical method based on the between-cluster entropy and its slight variation. Then, a novel procedure is presented, based on the calculation of the similarity between centroids, with successive refinement to allow effective identification of the outliers. The outcomes of the classification carried out by using the proposed procedure are compared to the results of other available techniques, using a set of clustering validity indicators for ranking the clustering methods. On the basis of these results, it emerges that the novel procedure exhibits better clustering performance with respect to both the literature approaches and the classical entropy-based method, for different numbers of clusters. The results obtained are of key relevance for assisting the electricity suppliers in identifying a reduced number of load pattern dependent classes, to be associated with distinct consumer groups for load aggregation or tariff purposes.
File in questo prodotto:
Non ci sono file associati a questo prodotto.
Pubblicazioni consigliate

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11583/2371562
 Attenzione

Attenzione! I dati visualizzati non sono stati sottoposti a validazione da parte dell'ateneo