Page 69 - MSAM-4-3

Materials Science in Additive Manufacturing Bead geometry prediction in laser-arc AM

A

B

Figure 10. Comparative analysis of various ensemble forecasting
approaches

A
Figure 12. Importance ranking of process parameters in the height
prediction task. (A) SHAP summary plot. (B) Importance plot of process
parameter
Abbreviations: f: Pulse correction; l: Arc length correction; p: Laser
power; v : Welding speed; v : Wire feed speed, SHAP: SHapley Additive
t
w
exPlanations
their effects may invert, thereby affecting the quality of
formation.
B
3.4. Conceptual framework for bead size
and shape prediction applied to in-process
characterization
Figure 14 depicts the initial stage of the additive
manufacturing process, in which the 3D model is first sliced
and a deposition path is generated; the slicing operation
defines the layer thickness, and the path planning sets the
track width and selects the appropriate process settings
accordingly. This study concentrates on the high-precision
Figure 11. Importance ranking of process parameters in the width
prediction task. (A) SHAP summary plot. (B) Importance plot of process prediction of bead geometry.
parameter Despite the high accuracy achieved in predicting
Abbreviations: f: Pulse correction; l: Arc length correction; p: Laser
power; v : Welding speed; v : Wire feed speed, SHAP: SHapley Additive bead morphology, variations in material batches and
t
w
exPlanations environmental fluctuations may still prevent the optimized
process parameters from faithfully reproducing the target
observed for bead width. Below 7 m/min, wire feed speed dimensions. Additional dimensional deviations can
has a negative influence, which reverses to positive when arise from tool wear, calibration errors, drifts in laser or
exceeding this value, facilitating height growth. Welding arc power, and fluctuations in wire-feed rate. To address
speed exhibits a positive correlation under 580 mm/min, these issues, we propose embedding the high-precision
shifting to a negative correlation beyond 610 mm/min, prediction model within an online quality-monitoring
thereby inhibiting height increase. Below 2.3 kW, laser and closed-loop control framework to enable real-time
power positively correlates with height, but the relationship correction and stabilization of bead dimensions.
reverses above this threshold, restricting height growth. In
summary, clear thresholds exist for process parameters During printing, a melt-pool camera continuously
affecting weld bead width and height, beyond which captures images near the welding torch and feeds them to

Volume 4 Issue 3 (2025) 12 doi: 10.36922/MSAM025220036