Optimization of Resistance Spot Welding parameters in steel sheets via Genetic Algorithm Implementation

This study quantitatively assesses the efficacy of Genetic Algorithm in the optimization of input process parameters of resistance spot welding. Two joints were fabricated (a) IFHS-IFHS of thickness 1.2 mm and (b) Mild steel- Mild steel of thickness 1.5 mm in a lap-shear joint arrangement using a single pulse welding. A D-optimal, Response surface methodology was used for designing and conducting the experiment. Regression models comprising of three continuous input variables (~ Weld-Time, Weld Current, Pressure) and one discrete variable (~ material type), with two responses (~ Failure load and Elongation) were used to formulate the input-output relationship. Henceforth, the developed regression models were used as fitness function in genetic algorithm optimization. The input variables were used as genetic codes for the chromosomes, which go through the process of selection, cross-over, mutation for the optimization process to complete. The sensitivity analysis of the Pareto-optimal scatter plots, (~ decision variables vs responses) were carried out. It was found that there was an increase of 5% and 6.6% in failure loads, with no substantial increase in elongation for both the IFHS and Mild steel joints respectively. The upper boundary limits of weld-time (~ 25 cycles), weld current (~ 15 KA) and pressure varying in the range [IFHS (~ 4.8–5.5 bar), Mild steel (~ 4.2–5.2 bar)] were found to be optimal input parameters for fabricating robust joints. This research provides a descriptive methodology of the application of genetic algorithm in the optimisation of process parameters for resistance spot welding process.