Implementasi Machine Learning pada Deteksi Kecacatan Printed Circuit Board

Ahnaf Zainul Muttaqin; Ibrahim Nawawi; Johan Pamungkas

doi:10.24036/voteteknika.v12i1.127224

Implementasi Machine Learning pada Deteksi Kecacatan Printed Circuit Board

Ahnaf Zainul Muttaqin⁽¹⁾, Ibrahim Nawawi⁽²⁾, Johan Pamungkas⁽³⁾,
(1) Tidar University

Indonesia
(2) Tidar University

Indonesia
(3) Tidar University

Indonesia

Corresponding Author

DOI : https://doi.org/10.24036/voteteknika.v12i1.127224

Full Text:

Language : id

Abstract

Kualitas suatu rangkaian elektronik sangat bergantung pada kualitas Printed Circuit Board (PCB) untuk mendukung proses berjalannya semua komponen. Meskipun dapat menggunakan inspeksi manual terhadap kecacatan pada PCB, namun metode tersebut memiliki keterbatasan terhadap ketelitian dan waktu. Oleh karena itu, penelitian ini dilakukan untuk mengimplementasikan Machine Learning dalam deteksi kecacatan pada PCB. Dalam pengaplikasiannya, penelitian ini menggunakan algoritma deteksi objek YOLOv5 (You Look Only Once versi 5), yang merupakan pengembangan dari convolutional neural network. Fokus penelitian ini adalah membangun sistem yang dapat mengidentifikasi kecacatan pada PCB menggunakan algoritma YOLOv5. Dataset yang digunakan terdiri dari 600 gambar, dengan 480 data latih dan 120 data uji. Jenis kecacatan PCB yang diidentifikasi meliputi open circuit, missing hole, mouse bite, short, spur, dan spurious copper. Hasil penelitian menunjukkan bahwa penggunaan YOLOv5m menghasilkan nilai mean Average Precision (mAP) sebesar 95,3%. Selain itu, dalam pengujian dengan 120 data uji, model berhasil mencapai akurasi sebesar 93,83%, presisi 98,13%, recall 95,43%, dan error 6,17%. Selain itu peningkatan spesifikasi hardware juga berpengaruh dalam kecepatan deteksi objek kecacatan pada PCB.

Kata kunci : PCB, kecacatan, Machine Learning, convolutional neural network, YOLOv5

The quality of an electronic circuit depends significantly on the quality of the Printed Circuit Board (PCB) supporting its operation. While manual inspection of defects on PCBs has been carried out, this method has limitations in terms of precision and time. Therefore, this research aims to implement Machine Learning in detecting defects on PCBs. In its application, this study utilizes the YOLOv5 (You Look Only Once version 5) object detection algorithm, which is an advancement of the convolutional neural network. The focus of this research is to build a system capable of identifying defects on PCBs using the YOLOv5 algorithm. The dataset used consists of 600 images, with 480 training data and 120 test data. Types of PCB defects identified include open circuit, missing hole, mouse bite, short, spur, and spurious copper. The research results indicate that using YOLOv5m yields a mean Average Precision (mAP) value of 95.3%. Additionally, in testing with 120 test data, the model achieved an accuracy of 93.81%, precision of 98.13%, recall of 95.43%, and an error rate of 6.17%. Furthermore, the improvement of hardware specifications also influences the speed of object detection for defects on PCBs.

Keywords: PCB, defect, Machine Learning, convolutional neural network, YOLOv5

References

S. Tang, F. He, X. Huang, and J. Yang, “Online PCB Defect Detector On A New PCB Defect Dataset,” Feb. 2019, [Online]. Available: http://arxiv.org/abs/1902.06197

I. C. Chen, R. C. Hwang, and H. C. Huang, “PCB Defect Detection Based on Deep Learning Algorithm,” Processes, vol. 11, no. 3, Mar. 2023, doi: 10.3390/pr11030775.

Y. Ningsih, D. Wahiddin, S. Arum, and P. Lestari, “Implemtasi Metode Canny Edge Detection Untuk Identifikasi Defect Solder,” vol. II, no. 1, 2021.

B. Farnham, S. Tokyo, B. Boston, F. Sebastopol, and T. Beijing, “Hands-on Machine Learning with Scikit-Learn, Keras, and TensorFlow Concepts, Tools, and Techniques to Build Intelligent Systems SECOND EDITION.”

U. Sri Rahmadhani and N. Lysbetti Marpaung, “Klasifikasi Jamur Berdasarkan Genus Dengan Menggunakan Metode CNN,” vol. 8, no. 2, 2023.

F. Paraijun et al., “Implementasi Algoritma Convolutional Neural Network Dalam Mengklasifikasi Kesegaran Buah Berdasarkan Citra Buah,” vol. 11, no. 1, 2022, doi: 10.33322/kilat.v11i1.1458.

R. Ding, L. Dai, G. Li, and H. Liu, “TDD-Net: A tiny defect detection network for printed circuit boards,” CAAI Trans Intell Technol, vol. 4, no. 2, pp. 110–116, Jun. 2019, doi: 10.1049/trit.2019.0019.

T. Mahendrakar et al., Performance Study of YOLOv5 and Faster R-CNN for Autonomous Navigation around Non-Cooperative Targets.

V. R. Joseph, “Optimal ratio for data splitting,” Stat Anal Data Min, vol. 15, no. 4, pp. 531–538, Aug. 2022, doi: 10.1002/sam.11583.

P. Kanah Arieska, N. Herdiani, P. Studi Ilmu Kesehatan Masyarakat, F. Kesehatan, and U. Nahdlatul Ulama Surabaya Alamat, “PEMILIHAN TEKNIK SAMPLING BERDASARKAN PERHITUNGAN EFISIENSI RELATIF,” 2018. [Online]. Available: http://jurnal.unimus.ac.id