For topology optimization problems involving large-scale structure features， such as blended wing body aircraft， traditional topology optimization methods face significant computational limitations when dealing with design domains on the order of millions of elements， making it difficult to obtain meaningful load path designs. To address this challenge， this paper proposed a topology optimization method based on reduced-dimensional models. By establishing mapping relationships between three-dimensional solid structures and their reduced-dimensional models， a hierarchical optimization framework is constructed. During the optimization iteration phase， the reduced-dimensional model is employed for optimization and design variable updates， while the mapping functions are used to transfer real-time optimization results to the three-dimensional solid model for high-fidelity mechanical response analysis. This forms a “reduced-dimensional optimization- three-dimensional validation” closed-loop feedback mechanism. Additionally， a size control function is introduced to enhance the manufacturing feasibility of the primary load-bearing structure. Through engineering case， the proposed dimensionality-reduction-based topology optimization method demonstrates its efficiency and stability advantages in solving high-dimensional optimization problems.