Litcius/Paper detail

Entire-Process Simulation of Friction Stir Welding — Part 2: Implementation of Neural Networks

American Welding Society, Yuming Xie, Xiangchen Meng, Yongxian Huang

2022Welding Journal19 citationsDOIOpen Access PDF

Abstract

To further understand the structure-parameter-property relationships of friction stir welded aluminum alloy joints, a nested neural network was proposed to map the macro- and microstructural response. The uncoupled effect of each primitive parameter on the joint performance was depicted. Reducing heat input and keeping an adequate load-bearing area of the welding nugget zone were proven to be the sufficient and necessary conditions to obtain high load-bearing performance. The entire-process simulation strategy showed great potential for prediction and optimization of the macro- and microstructural response of complex and large components.

Topics & Concepts

WeldingFriction stir weldingMacroJoint (building)Load bearingProcess (computing)Structural engineeringBearing (navigation)Artificial neural networkMaterials scienceMechanical engineeringAlloyAluminiumProperty (philosophy)MetallurgyEngineeringComputer scienceOperating systemEpistemologyProgramming languageMachine learningArtificial intelligencePhilosophyAdvanced Welding Techniques AnalysisWelding Techniques and Residual StressesMetal Forming Simulation Techniques
Entire-Process Simulation of Friction Stir Welding — Part 2: Implementation of Neural Networks | Litcius