Machine learning techniques in monitoring and controlling friction stir welding process: a critical review
Abstract The friction stir welding (FSW) technique is a solid-state joining method that presents numerous advantages compared to conventional welding techniques, including enhanced mechanical joint efficiency and the ability to join challenging materials such as aluminium and magnesium. FSW is a sop...
Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Springer
2025-05-01
|
| Series: | Discover Applied Sciences |
| Subjects: | |
| Online Access: | https://doi.org/10.1007/s42452-025-07133-8 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Abstract The friction stir welding (FSW) technique is a solid-state joining method that presents numerous advantages compared to conventional welding techniques, including enhanced mechanical joint efficiency and the ability to join challenging materials such as aluminium and magnesium. FSW is a sophisticated process characterized by the interaction of various parameters during welding. Inadequate selection of these parameters can lead to defects such as void, tunnel, flash, and other surface and subsurface issues, tool failure, and compromised joint strength. The quality of the weld is essential for ensuring the durability and safety of the final product. Establishing optimal process parameters requires extensive trial and error, hindering productivity. This challenge can be effectively addressed by leveraging recent advancements in the manufacturing industry, particularly the 'Machine Learning: Industry 4.0 Revolution' approach. Machine learning aims to enhance outcomes by integrating digital and physical processes, facilitating efficient monitoring and management of machine performance. By employing machine learning techniques, the FSW process can become more cost-effective through optimizing process parameters, early detection of defects and tool failures, reduction of waste, and attainment of superior joint properties, all while minimizing the need for extensive trial and error. This review article critically explores how machine learning can optimize process parameters, identify and rectify defects, predict and manage tool failures, and control corrosion rates to achieve high-quality FSW joints.Using causative variable model accuracy in predicting tool failure increases from 87 to 98.1% as compared to raw FSW process variables. |
|---|---|
| ISSN: | 3004-9261 |