Mineração de Dados Aplicada à Identificação de Fontes de Correntes Harmônicas em Consumidores Residenciais (Data Mining Applied to Identification of Harmonic Sources in Residential Consumers)

Ricardo Augusto Souza Fernandes (ricardo.asf@gmail.com)1, Ivan Nunes Silva (insilva@sc.usp.br)1, Mario Oleskovicz (insilva@sc.usp.br)2


1Universidade de São Paulo
2olesk@sc.usp.br

This paper appears in: Revista IEEE América Latina

Publication Date: June 2011
Volume: 9,   Issue: 3 
ISSN: 1548-0992


Abstract:
This work presents a promising method to identify harmonic current sources commonly encountered in residential electrical systems. From this method, feasible solutions can be applied to mitigate the high levels of harmonic currents, generated by nonlinear loads. Techniques of feature selection were used to data preprocessing and to minimize the effort in identification of loads connected to the electrical system. This process was compound to artificial neural networks, which attempt map the system dynamics in order to obtain general solutions. All harmonic distortion situations were created in laboratory from a power source, and in its outputs were inserted the loads and power quality analyzers, which perform the extraction of all measurements. These signals were then preprocessed and used to training and validating of the neural structures. The obtained results were considered satisfactory, which confirm the use of this method to solve many problems related to disturbances caused due to bad power quality.

Index Terms:
Cargas não lineares, componentes harmônicas, identificação de fontes harmônicas, redes neurais artificiais   


Documents that cite this document
This function is not implemented yet.


[PDF Full-Text (535)]