Construção de um banco de dados simulado de imagens de isoladores utilizados em cabos de alta tensão

Abstract

String insulators are a series of insulating pieces that are placed on an electrical power transmission line and are designed to withstand mechanical and electrical load, as well as to insulate the electrical line from contact with metallic parts of the tower. String insulators must be cleaned regularly to ensure they are working properly, as different types of compounds can build up on the insulators, affecting their function. The verification of the need for cleaning is done by visual observation, which can cause false statements. Algorithms that perform this type of classification need a considerable dataset to perform their tasks accurately and efficiently. In the literature, the existing images are limited, mainly in the case of dirty insulators. Therefore, this work aims to develop a simulated dataset, using software such as SolidWorks, Inventor and Unity 3D and with other existing images on the internet and unknown real ones in the field.