DETECTION OF FAULTY ISOLATORS IN ENERGY TRANSMISSION LINE BY USING CONVOLUTIONAL NEURAL NETWORKS


Noğay H. S.

5th INTERNATIONAL CONGRESS ON ENGINEERING SCIENCES AND MULTIDISCIPLINARY APPROACHES, İstanbul, Turkey, 25 - 26 February 2023, pp.282-287

  • Publication Type: Conference Paper / Full Text
  • City: İstanbul
  • Country: Turkey
  • Page Numbers: pp.282-287
  • Bursa Uludag University Affiliated: Yes

Abstract

Abstract: Electrical energy transmission line insulators are one of the important elements used at connection points in overhead transmission lines and whose main task is insulation. Since they generally do not require maintenance, they need to be replaced immediately when they are damaged due to weather conditions or any reason, while their replacement is delayed because they are not followed by a tracking system. Progressive preliminary studies are needed to develop automatic detection and tracking systems. In this context, there is a need to implement and design a system that can automatically detect damaged isolators from image data using artificial intelligence. In order to meet this need, a CNN model based on isolator images was designed and tested in this study. As a result of the test, the average prediction value was reached with an accuracy of 89%, and the defective insulators were determined as completely correct. Keywords: Damaged Insulators, Deep Learning, CNN, TL