Owing to the increasing demand for raising the thrust weight ratio of modern aero engine,it is very important to optimize the structures of the components.Traditional optimization methods of structure design are time consuming and hard to be put into practice.So,in this paper,a new method of structure design optimization is induced to which both BP neural networks and genetic algorithms (in short:BPN GA) are applied.A program which contains 9 variables is designed for the structure optimization of a disk model with the BPN GA method,which proves that it has better calculating rate and precision than those with traditional optimization methods.