Quasi-quantum computing for constraint-aware optimization of laser welding parameters in aluminum alloys
Year: 2026
Type: article
Abstract: Laser welding of aluminum alloys used in the automotive and aerospace industries requires high-quality joints through optimization of high-dimensional process parameters, such as laser power, welding speed, focus offset, and wobbling patterns, under practical constraints. However, conventional trial-and-error and gradient-based approaches are inefficient in such a large parameter space. In this study, we formulate the problem as an Ising model and propose a quasi-quantum computing framework for constraint-aware optimization using the Fujitsu Digital Annealer. A machine-learning model for joint width prediction is first built and then approximated as an Ising-type AI model. Practical constraints, including alloy combinations, sheet thickness, and processing conditions, are explicitly incorporated to obtain feasible solutions. Experimental lap welding validation confirms good agreement between predicted and measured values and achievement of the target joint width. The proposed approach also reduces search time by several hundredfold, enabling rapid and reliable laser welding process design.
Source: MRS Advances
Institution: Resonac (Japan)
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: hybrid
Estimated APC paid: $3,390
OpenAlex ID: W7226026427
