Overview and comparison of neural network models for diagnostics and quality control in the production of printed circuit boards

• Дата: 2025
• Автори: Верлань , А.А., Фастовець , Є.Р.
• Формат: Стаття
• Мова: English
• Опубліковано: Інститут проблем реєстрації інформації НАН України 2025
• Теми: quality control; defect detection; neural networks; artificial intelligence; printed circuit boards; YOLO; optical inspection
• Онлайн доступ: http://drsp.ipri.kiev.ua/article/view/336051
• Назва журналу: Data Recording, Storage & Processing

The article reviews and analyzes modern solutions based on neural networks for diagnostics and quality control of printed circuit boards (PCB). Various convolutional neural network (CNN) models, including LeNet-5, AlexNet, VGGNet, ResNet, and others, are reviewed, and their advantages and disadvantages are discussed. The YOLO (You Only Look Once) model, which is capable of detecting objects in real time at high speed, is considered in more detail. First part gives a general overview of neural networks and their work in the context of working with visual data. Next, the article addresses individual CNN models and their use in the context of PCB diagnostics and control. The work provides a comparative analysis of actions and results of application of neural networks and traditional testing methods along with the challenges and prospects.