Yazar "Al-azzawi, Ibraheem Ismael Ibraheem" seçeneğine göre listele

Listeleniyor 1 - 1 / 1
  • Küçük Resim
    Yayın
    Artificial intelligence utilization in production quality management: Piping fabrication
    (İstanbul Gedik Üniversitesi, 2021) Al-azzawi, Ibraheem Ismael Ibraheem; Demir, Arif
    The development of quality control systems in production is one of the matters of concern to producers and quality control institutions alike. The pipe industry is considered one of the important industries because of its wide and sensitive applications in the infrastructure of projects, so limiting quality control to human capabilities only is considered a weakness in these systems. The benefit of the applications of artificial intelligence in quality control has been used by many producers, despite the difficulties that accompany the use of these techniques, as they require training and a special work environment. Deep neural networks were used in this work to monitor the quality of tubes and classify them according to the criteria that were determined. One of the things that have been focused on is to reduce the time and data needed to train the deep game network, by relying on the technology of transfer learning. Transfer learning is one of the ways to train deep neural networks that have been previously trained, which saves time and data for training and allows the user to use high-resolution models in the design and make some modifications to them, making them commensurate with the nature of the application he wants to design. Alexie is used in this work because it has high accuracy in the classification of images. Alexie contains 25 layers, only thee laters were modified to be suitable with our application. About 1000 images were only used in training and validation processes. The simulation results show that the transfer learning is an accepted technique and is also saves time with high accuracy results.