Intelligent Multi-objective Optimization Design for Aerodynamic Layout of Mechanical Expansion Reentry Vehicle

Aerodynamic configuration optimization is a key technology in aircraft design. As a complex system, the aerodynamic layout optimization of mechanical expansion reentry vehicle involves several optimization objectives, such as deceleration effect, heat protection effect and space utilization ratio. Aiming at the problem of large computation in computational fluid dynamics (CFD) optimization of reentry vehicle, an intelligent multi-objective optimization method based on BP neural network is proposed. Firstly, the shape of the reentry vehicle is parameterized. The optimal Latin hypercube experimental design is used to generate sample points, and the high precision aerodynamic and thermal performance calculation is carried out by CFD method to obtain the drag coefficient and maximum surface heat flux. BP neural network is adopted to non-linear fit the sample set, and the approximate model of neural network was constructed. NSGA-II algorithm was used for multi-objective optimization of three objective functions. Pareto solution set and frontier with good distribution were obtained, the variation rule between parameters and objectives was explored by sensitive analysis. The results show that the Intelligent approximate model can quickly solve the optimization problem under the premise of ensuring the accuracy, which provides a reference for future aircraft design and application.