<p>To address the limitations of conventional truck frame optimization—specifically, the inadequate integration of material selection and structural parameter design as well as the limited efficiency and accuracy of performance prediction—this study proposes a structure–material–performance integrated intelligent optimization framework. A high-fidelity simulation platform was developed through finite element modeling and experimental validation, enabling the evaluation of the frame’s static performance and fatigue life. In this context, the material type was innovatively introduced as a design variable, alongside dimensional and geometric parameters, to construct a coupled multi-objective optimization model that captures the interdependence among structural configuration, material selection, and performance requirements. To enhance optimization efficiency and predictive accuracy, a multi-task deep neural network surrogate model based on the wide and deep architecture was developed, enabling unified prediction of 15 key performance metrics. The NSGA-III algorithm was adopted to simultaneously optimize structural mass, torsional displacement, and modal frequency, subject to constraints on maximum stress and fatigue life, resulting in a Pareto-optimal solution set. Subsequently, a hybrid CRITIC–AHP weighting method integrated with the VIKOR decision-making approach was employed to determine the optimal trade-off among multiple objectives. The results demonstrate that the optimized design achieves a 13.46% reduction in structural mass compared to the original configuration, without compromising strength or stiffness. The proposed framework effectively overcomes traditional decoupling between material selection and structural parameter optimization, significantly enhancing the multi-performance collaborative optimization capability for complex structures and exhibiting strong potential for practical engineering applications.</p>