Prediction of Electron Beam Welding Penetration Depth Using Machine Learning-Enhanced Computational Fluid Dynamics...

Area of Science:

• Materials Science and Engineering
• Computational Physics
• Manufacturing Technology

Background:

• Precise prediction of electron beam welding (EBW) penetration depth is crucial for quality control.
• Traditional methods like regression analysis and neural networks can be time-consuming and expensive.
• Existing predictive models often lack efficiency and require extensive preliminary testing.

Purpose of the Study:

• To develop a novel, efficient, and accurate approach for predicting EBW penetration depth.
• To integrate computational fluid dynamics (CFD) modelling with artificial neural networks (ANN) for enhanced predictive accuracy.
• To reduce the time and financial resources required for EBW process optimization.

Main Methods:

• Synergistic combination of computational fluid dynamics (CFD) modelling and artificial neural networks (ANN).
• Application of CFD for simulating the physical processes involved in EBW.
• Training and validation of ANN models using CFD-generated data and experimental parameters.
• Development of a beam characterisation method for broader applicability.

Main Results:

• The integrated CFD-ANN approach achieved high accuracy in predicting EBW penetration depth, with an average absolute percentage deviation of approximately 8%.
• Accurate predictions were consistent across a linear electron beam power range of 86 J/mm to 324 J/mm.
• The method significantly reduces the need for costly and time-intensive preliminary testing.

Conclusions:

• The combined CFD-ANN model offers an efficient and accurate solution for predicting EBW penetration depth.
• This approach enhances control over EBW weld quality by enabling fine-tuning of key process variables.
• The methodology is adaptable to different electron beam machines, offering versatile application in industrial settings.