About this Abstract |
Meeting |
2021 AWS Professional Program
|
Symposium
|
2021 AWS Professional Program
|
Presentation Title |
A Machine Learning Approach to Predict Weld Anomaly in WAAM and Welding Process |
Author(s) |
Yongzhe Fa, Xinghua Yu, Wei Ya, Remco Rook |
On-Site Speaker (Planned) |
Yongzhe Fa |
Abstract Scope |
Weld or WAAM defects detection usually requires post-weld inspection. As a result, wrong welding parameters or machine malfunction which induced those defects could not be known until the process was finished. Realtime defect detection and process control become critical to make defect-free welds. Current study established a number of machine learning models, such as the support vector machine, random forest, gradient boosting decision tree, LightGBM for anomaly detection. In addition, multiple mathematical statistical detection models are established according to the time sequence characteristics of data during the welding process, such as ARMA, ARIMA, as well as deep learning detection model LSTM. The result shows that the anomaly detection models established herein can detect welding defects and provide feedback to welding process control. |
Proceedings Inclusion? |
Definite: Other |