Investigating the effect of residual stresses and distortion of laser welded joints for automobile chassis and optimizing weld parameters using random forest based grey wolf optimizer

Abstract

The present investigation analyses the selection of the right welding method and joint and advanced testing methods (NDT) for highly durable automotive frames. Moreover, the present investigation analysis suggests the best machine learning (ML) algorithm for selecting the best weld method and optimal solution. The experiment was performed based on the response surface methodology (RSM) based design of the experimental approach. As a result, laser beam welding (LBM) and cross joint are the significant weld methods for automotive frames. The proposed ML algorithm successfully optimized the LBM input parameters as laser power = 1277 W, welding speed (WS) = 32.2 mm/s, focal point: 1 mm and working angle = 0.14 Rad with an average error of approximately 0.033. Based on the results, the optimum output weld parameters are bead width = 4322.7 µm, penetration depth (PD) = 3157.9 µm, total strain = 0.0098 mm/mm and residual stress = 645.2340 MPa, respectively.

Keywords:

• Laser beam welding
• random forest algorithm
• grey wolf optimization
• response surface methodology
• residual stresses

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

Data sharing does not apply to this article.