About this Abstract |
Meeting |
Advances in Welding and Additive Manufacturing Research 2022
|
Symposium
|
Advances in Welding and Additive Manufacturing Research 2022
|
Presentation Title |
Machine Learning Approach for Predicting Performance Properties of Al-steel Resistance Spot Welds |
Author(s) |
Wei Zhang, Dali Wang, Jian Chen, Hassan Ghassemi-Armaki, Blair Carlson, Zhili Feng |
On-Site Speaker (Planned) |
Wei Zhang |
Abstract Scope |
Automakers are dedicated to advancing core technologies to reduce the vehicle's energy consumption and CO2 emissions. A key technology area is related to manufacturing of vehicle body structure components by joining high-strength steel with lightweight materials, such as advanced high strength steels with aluminum alloys. Resistance spot welding (RSW) has been as a dominant joining method for automotive body assembly. In reality, making robust dissimilar material RSWs of the aluminum with steel alloys is still challenging because of the great differences in physical properties and metallurgical incompatibility of the two alloys. This study explores the feasibility of using machine learning (ML) methods to elicit the process-structure-property relationship at an accelerated pace which is important for manufacturing joints with desired performance and high quality. A supervised algorithm of deep neural network (DNN) was applied to analyze the post processed joint data with an emphasis on the weld structure/attribute and property relationships. Data being assessed comprised of material specifications, weld conditions, microstructural features, weld attributes, and bulk mechanical performance properties. The trained DNN is capable to identify the high dimensional relationships between weld structures/attributes and joint properties and to predict the joint performance properties with good accuracy. The DNN for RSW is an extensible model with unified architecture, which is expendable to additional data streams from different materials, thickness, and other combinations, serving as a flexible and cost-effective tool for RSW quality assurance. |
Proceedings Inclusion? |
Undecided |