Modeling and optimization of conflicting responses in the laser transmission welding process using RSM, PSO, and TLBO algorithm

Abstract

Laser transmission welding (LTW) is a widely used polymer welding technology in industries today. The performance of LTW is governed by a number of process parameters, and fine-tuning those parameters is critical for the process to achieve the intended results. Optimization of process parameters for enhancing LTW performance, like weld strength and weld width, is always tricky since they are inherently contradictory. Therefore, in this research, an effort is made to optimize the LTW parameters utilizing the three best-known optimization approaches, and the performance of those optimization approaches in LTW process optimization is compared. First, the response surface method (RSM) is employed to build empirical equations that establish an empirical correlation between process variables and desired performance features. These empirical equations are then employed as objective functions for process optimization utilizing the RSM-based desirability function approach (DFA), particle swarm optimization (PSO), and teaching learning-based optimization (TLBO) algorithms. The performance of the chosen optimization approaches is compared with reference to optimum results, accuracy, convergence rate, and computing time. PSO and TLBO algorithms outperform the DFA approach for single and multi-objective optimizations. TLBO is found to have faster convergence, whereas PSO takes less time for computation.