Sub-Project 8(Yeong-Do Park): Development of a Machine Learning-based quality prediction and management system for welding and joining processes in electric vehicle applitcations
Key Research Activities
Collect joining quality data according to various steel grades and plate thickness combinations, build a database, and derive the optimal process to improve joining quality
Establish new standardized criteria for SPR (Self-Piercing Riveting) and REW (Resistance Element Welding) quality evaluation, and conduct research for international standard certification to meet global automotive industry demands
Quantify the bonding quality of REW and SPR processes and label the data based on the presence or absence of weld defects
Develop predictive technology capable of simultaneously detecting surface and internal defects through non-destructive testing
Extract key indicators from real-time monitoring data and predict joint integrity by analyzing correlations between process variables and weld quality
Design AI/ML-based models and algorithms to monitor and predict quality variations in real-time during SPR and REW joining processes
Research Description
Collect and build a database of joint quality data for various steel types and sheet thickness combinations, and optimize processes to improve joint quality
Establish new standardized criteria for evaluating SPR and REW joint quality, and conduct research based on international standard certifications to meet the needs of the global automotive industry
Quantify the joint quality results of REW and SPR processes and label them according to the presence or absence of defects
Develop predictive technologies capable of detecting both surface and internal defects simultaneously through non-destructive inspection
Extract key indicators and predict joint integrity by analyzing correlations between welding quality and process variables based on real-time monitoring data
Develop a joint quality prediction model using AI/ML (Artificial Intelligence/Machine Learning), and design algorithms to monitor and predict quality variations occurring in SPR and REW processes in real time
