Abstract: The lightweight structure can effectively reduce the energy consumption of the whole vehicle and enhance its dynamic performance, and aluminum alloy has been used in the design of related components due to its lightweight and high-strength characteristics. Arc additive manufacturing has advantages such as low cost and high efficiency, which can be used for the personalized customized preparation of components, but its excessively high heat input leads to low forming accuracy. To address this issue, additive manufacturing experiments on 5356 aluminum alloy were conducted based on three welding technologies: melt inert-gas welding additive manufacturing (MIG-AM), double-pulse melt inert-gas welding additive manufacturing (DP-MIG-AM), and cold metal transfer additive manufacturing (CMT-AM). The influence of different combinations of three process parameters, namely wire stick-out length, welding current, and welding speed, on the macroscopic morphology of single-pass and single-layer weld beads was investigated. Based on the macroscopic geometric feature data collected from the single-pass and single-layer welds, the qualitative and quantitative relationships among parameters, quality, and forming were established. A data-driven machine learning method was used to predict the geometric morphological characteristics of the weld seams. Furthermore, a particle swarm optimization algorithm was introduced to optimize the model to improve the iteration rate and accuracy. Finally, relatively optimal process parameters were selected for the three welding technologies respectively to conduct single-pass and multi-layer experiments, and the deposition efficiency and mechanical properties were evaluated.