Development of artificial intelligence algorithms for detection of defects in surge arresters from thermal images
Guardado en:
Autores: | , , |
---|---|
Formato: | artículo original |
Fecha de Publicación: | 2021 |
Descripción: | The main faults in surge arresters are commonly associated with current rises and, therefore, with heating by the Joule effect. Thus, thermovision is a very suitable technique for detecting defects in this electrical power system equipment. However, the use of thermovision to identify possible failures in surge arresters depends on an experienced operator to interpret the obtained results, which many times are subject to interpretation errors. This work aims to apply Artificial Intelligence techniques to classify surge arrester thermal images. Artificial Intelligence algorithms based on Artificial Neural Networks and deep learning techniques were developed for this purpose, considering that many authors have succeeded in using these methods in failure diagnosis in other equipment in the electrical power system. The results obtained showed that the neural networks developed by the backpropagation algorithm presented good efficiency when classifying surge arrester thermal images, and there is no need to segment the surge arrester from the thermal image to carry out the classification. |
País: | RepositorioTEC |
Institución: | Instituto Tecnológico de Costa Rica |
Repositorio: | RepositorioTEC |
Lenguaje: | Español |
OAI Identifier: | oai:repositoriotec.tec.ac.cr:2238/13637 |
Acceso en línea: | https://revistas.tec.ac.cr/index.php/tec_marcha/article/view/6043 https://hdl.handle.net/2238/13637 |
Palabra clave: | Surge arresters thermal images backpropagation artificial neural networks classification Para-raios imagens termográficas redes neurais artificiais classificação |