Society's great and growing demand for buildings and structures has created the need to apply new construction methods to shorten construction times, make buildings lighter, extend their useful life, and make them more earthquake-proof. In the long term, the new methods will lead to structural optimization, increased building performance, and the achievement of optimal operating conditions. New technologies are meeting society's increasing need for special structures more than ever. Additive manufacturing is based on gas metal arc welding as one of the fastest and most cost-effective manufacturing methods for primary metal structures. For this purpose, the three parameters voltage, wire feed speed, and welding speed were considered initial parameters affecting the width and height of the welding flux. To investigate the effects of the process,

16 experiments with input parameters were evaluated. The width and height of the sweat pollen were determined by experimental investigations. Subsequently, the resulting welding geometry is modeled using three numerical modeling methods, including intensive learning machines, relevence vector machine, and fuzzy logic. The comparison between the experimental data and the results of the three generated models shows that fuzzy logic comes closest to the experimental data of the welding geometry of the modeling methods. For example, the test data of the generative fuzzy model resulted in an average error for height and width of 0.667 and 0.5477, respectively, and a root mean square error for height and width of 0.0046 and 0.3, respectively, which expresses the generalization ability and reliability compared to other modeling methods in this process. Finally, a metal pattern of a special structure was produced based on arc and wire additive manufacturing based gas metal arc welding.