DocumentCode
3495871
Title
Efficient encoding of customer class load profiles
Author
Beretka, Sandor F. ; Varga, Ervin D.
Author_Institution
Fac. of Tech. Sci., Univ. of Novi Sad, Novi Sad, Serbia
fYear
2013
fDate
9-12 Sept. 2013
Firstpage
1
Lastpage
5
Abstract
The majority of distribution management functionalities rely on load profiles. Customer classification and load analysis have the largest impact on them. In this paper a novel approach for load profile generation is presented. The presented work is based on artificial neural networks: sparse autoencoders and deep belief networks in order to reveal hidden features from data sets.
Keywords
belief networks; encoding; neural nets; power engineering computing; power system management; artificial neural networks; customer classification; deep belief networks; distribution management; encoding; load analysis; load profile generation; sparse autoencoders; Biological neural networks; Encoding; Feature extraction; Neurons; Training; Vectors; autoencoder; classification; load profile; neural network;
fLanguage
English
Publisher
ieee
Conference_Titel
AFRICON, 2013
Conference_Location
Pointe-Aux-Piments
ISSN
2153-0025
Print_ISBN
978-1-4673-5940-5
Type
conf
DOI
10.1109/AFRCON.2013.6757767
Filename
6757767
Link To Document