Optimizing Process Parameters in Laser Transmission Welding Solid PC/Porous-PET Using Prediction Models: Experimental...

Area of Science:

• Materials Science and Engineering
• Manufacturing Processes
• Computational Modeling

Background:

• Laser transmission welding (LTW) of dissimilar materials like solid polycarbonate (PC) and porous polyethylene terephthalate (porous-PET) presents significant challenges in parameter optimization.
• The complex interplay of factors influencing weld quality necessitates advanced modeling techniques.

Purpose of the Study:

• To establish a robust relationship between LTW process parameters and welding quality for solid/porous material combinations.
• To compare the efficiency of different optimization algorithms (GA, BO, CMA-ES) for determining optimal LTW parameters.

Main Methods:

• A comprehensive experimental dataset was generated using a flexible factor-level design, including welding power, speed, PC thickness, and porous-PET density.
• A Gaussian process regression (GPR) model was developed and optimized for high predictive performance.
• Genetic Algorithm (GA), Bayesian Optimization (BO), and Covariance Matrix Adaptation Evolution Strategy (CMA-ES) were employed for parameter optimization.

Main Results:

• The optimized GPR model trained on the full dataset significantly outperformed models trained on averaged data.
• CMA-ES exhibited the fastest convergence and shortest runtime among the tested optimization algorithms, achieving results comparable to GA and BO.
• Experimental validation confirmed the accuracy of the optimized parameters, resulting in uniformly formed weld seams with low relative error.

Conclusions:

• The proposed strategy, integrating GPR with CMA-ES, provides an efficient method for optimizing LTW of solid/porous materials.
• This approach offers significant potential for enhancing joint performance, improving process efficiency, and reducing manufacturing costs in the joining of advanced materials.