AUTOMATIC CLASSIFICATION OF HIGH VOLTAGE BREAKERS WITH DEEP LEARNING

Creative Commons License

Noğay H. S.

9 INTERNATIONAL CONGRESS ON ENGINEERING AND TECHNOLOGY MANAGEMENT, İstanbul, Türkiye, 29 Nisan - 01 Mayıs 2023, ss.174-181

  • Yayın Türü: Bildiri / Tam Metin Bildiri
  • Basıldığı Şehir: İstanbul
  • Basıldığı Ülke: Türkiye
  • Sayfa Sayıları: ss.174-181
  • Bursa Uludağ Üniversitesi Adresli: Evet

Özet

Abstract: High Voltage Breakers are critical in power systems, protecting the system against faults and overloads. By automatically classifying surge breakers, we can make sure they are working properly and identify potential problems before they become major problems. Deep learning methods can provide accurate and efficient classification, allowing faster and more reliable maintenance and repair of these critical components. In this study, a convolutional neural network model application that can perform multiple automatic classification of high voltage breakers is implemented. For this, the last three layers of the ResNet-18 pre-trained model were revised and used through the transfer learning technique. As a result of the study, the quad classification with the proposed method has reached 82% accuracy, and it has been understood that the proposed method can be used effectively in automatic classification by separating high voltage circuit breakers into different classes.