Date of Publication
Bachelor of Science in Computer Engineering
Gokongwei College of Engineering
Electronics And Communications Engg
Reggie C. Gustilo
Defense Panel Chair
Roderick Y. Yap
Defense Panel Member
Leonard U. Ambata
Melvin K. Cabatuan
Printed circuit boards (PCBs) are important components of electronics. They serve as the "heart" of any electronic device by connecting all electronic components. However, one of the challenges faced by manufacturers is the presence of defects on the boards during etching, which may render the board unusable. In the past, these defects were checked manually by manufacturers, which was very time consuming and difficult, especially if the PCB is very complex. Today, different ways of how to detect these defects are being proposed. Convolutional neural network (CNN), a deep learning algorithm optimized for image processing due to its flexibility and efficiency, is proposed to be used in PCB defect detection. The authors present this method for detecting etching defects, namely open lines and shorted connections, enclosing them in bounding boxes for detection. The proposed solution proved to be able to create an 85% accuracy CNN model which can predict the possible defects in a given PCB image via mobile phone. The model was then compared to previous solutions to determine whether the proposed solution was effective or not.
Keywords— PCB Fault Detection, Tensorflow, Image Processing, Convolutional Neural Network, Transfer Learning.
Chua, A. P., & Ong, D. Y. (2022). PCB fault detection through the use of convolutional neural networks. Retrieved from https://animorepository.dlsu.edu.ph/etdb_ece/13
Upload Full Text