Development of a hybrid ANFIS method for optimising laser beam welding of dissimilar metals: a virtual prototype approach

Laser beam welding (LBW) is a high-quality fusion joining process that has steadily gained popularity across various sectors due to its cutting-edge capabilities. The welding of dissimilar materials, such as Ti-6Al-4 V and Inconel 625, is particularly crucial in the aerospace industry due to the enhanced strength and low corrosion rates of these materials. To optimise the LBW process, experiments were designed using Taguchi’s method, varying parameters like laser power, welding speed, and pulse duration. Optimal conditions for minimising top and bottom weld width were identified as 2.5 kW LP, 3 mm/min WS, and 6.2 ms PD, while 3.5 kW LP, 2 mm/min WS, and 8.2 ms PD were found to maximise penetration and joint quality. The study revealed laser power as the most significant factor affecting weld dimensions. A Grey-based Adaptive Neuro-Fuzzy Inference System (ANFIS) was developed to create a reliable virtual prototype model for predicting LBW outcomes. The comparison of experimental results with model predictions showed a strong correlation, validating the model’s efficacy. The advanced AI-powered decision-making tool demonstrated robust performance in simulating LBW processes, enabling manufacturers to make more informed decisions. This research underscores the potential of virtual simulators in optimising design and manufacturing processes, promoting innovative and user-centred solutions in the fields of advanced manufacturing and engineering.