Traditional genetic algorithm has defects such as low optimization efficiency and low precision. We improved this algorithm by compiling a structural optimization program through Fortran language. The improved genetic algorithm was applied to the optimization of the design for a large-scale prestressed rectangular aqueduct. By comparing the original design with the optimized design in terms of safety and economy, we verified the feasibility of the optimized scheme. It’s proved that the optimized algorithm improves the convergence rate and accuracy of the solution. In the optimized design, each part of the aqueduct meets the stress and strain condition, and the three-dimensional mechanical behavior of the structure is more reasonable. The results demonstrate that the structural optimization reduces the construction costs and improves the structural behavior.