@Article{cmc.2026.078762,
AUTHOR = {Chin Ta Wu, Shing Han Li, Ching Shih Tsou},
TITLE = {Integrating FDC and Machine Learning for Enhanced Anomaly Detection in WB Bonding Joint Quality},
JOURNAL = {Computers, Materials \& Continua},
VOLUME = {},
YEAR = {},
NUMBER = {},
PAGES = {{pages}},
URL = {http://www.techscience.com/cmc/online/detail/26707},
ISSN = {1546-2226},
ABSTRACT = {In semiconductor packaging processes, the wire bonding procedure, which connects chips to substrate lead frames using metal wires, is a crucial step. The quality of the bonding joints significantly affects product performance, including signal integrity and reliability, and is challenging to verify after subsequent processes. To mitigate the risk of defective bonding joints entering the assembly packaging stages of production, this study integrates the concepts of Fault Detection and Classification (FDC) and machine learning into the wire bonding process for enhanced anomaly detection. Production data from the machines were collected and analyzed using statistical methods to filter out normal bonding joint data. After conducting feature engineering, we developed an anomaly detection model specifically for bonding joints. This inference model was subsequently deployed and validated using actual production data. During the validation phase, the proposed anomaly detection system effectively assisted the production line in identifying ball-related anomalies, thereby preventing these defects from advancing to later stages and ensuring overall product quality.},
DOI = {10.32604/cmc.2026.078762}
}