ALBUQUERQUE, E. L.; http://lattes.cnpq.br/0965589476872239; ALBUQUERQUE, Eduardo Lopes de.
Résumé:
The present study sought to develop a tool for diagnosis of air pollution in
insulators string using artificial intelligence and electronic counters partial discharge
sensors installed on transmission lines in Northeastern Brazil. Laboratory experiments
to generate data about the behavior of leakage current in insulators submitted to five
different known levels and pollution were used. With this data was trained an artificial
neural network - ANN in search of creating a classifier that could receive as input the
information collected by electronic sensors and provides as output the degree of
pollution of the insulator without it being necessary to remove the isolators from its
activity in the field nor of human presence or depend near the insulator under study,
since the data is sent via satellite by electronic sensors. The neural network showed no
percentage desired for immediate implementation of the project hit, but this work
provides important benchmarks below to advance the idea proposed in this research will
bring huge gains in standardizing diagnostic pollution on insulators, reducing costs with
labor to inspection work, better working conditions for transmission lines inspectors
(because do not need to spend all night in dangerous conditions in search of polluted
insulators) and ensuring greater reliability to the electric power system with continuous,
uninterrupted monitoring system sensors.