<p>In the context of automotive lightweight strategies, the use of steel/aluminum welded structures in place of pure steel is of significant importance for reducing vehicle weight. By adjusting welding process parameters and adding an interlayer, the formation of intermetallic compounds (IMCs) can be effectively reduced, thereby improving the quality of the welded joints. This paper establishes a CatBoost model to describe the nonlinear relationship between welding process parameters, the addition amount of CeO₂, and optimization objectives. The robust iterative multiview embedding (RIME) is employed for hyperparameter tuning, enhancing the accuracy of the algorithm's predictions. Based on the predictions of the RIME-CatBoost model and the calculated welding costs, the multi-objective african vulture optimization algorithm (MOAVOA) is used for optimization, and the CRITIC comprehensive evaluation method is applied to find the optimal parameter combination. Validation analysis shows that, compared to pre-optimization experimental values, the deformation of the optimized welded samples decreased by 37.7%, the maximum tensile force increased by 20.6%, and the cost was reduced by 18%, demonstrating the feasibility of the proposed method.</p>