Pendeteksian Cacat Isolator pada Jaringan Listrik Menggunakan Convolutional Neural Network yang Terkoneksi dengan Kamera

Abstract

Insulators on the power grid play an important role in maintaining the reliability and safety of the power system, including separating electrical conductors from the supporting structure. This is important to prevent short circuits or leakage currents between conductors and support structures that can cause interference with the power grid system. To facilitate the inspection of insulators on the power grid, a prototype system is made using the Convolutional Neural Network method which is programmed by training on images of a number of insulators to detect whether the insulator is still in good condition (normal) or already in a state of defect. The Convolutional Neural Network system or program that has been made successfully detects insulator objects and provides an assessment of whether the insulator is in good condition (normal) with an accuracy level of 0.43 or in a defective state with an accuracy level of 0.49