Configuration Optimization of Reentry Capsule Based on Surrogate Model

This paper investigates the multi-object optimization design of reentry capsule configuration under typical hypersonic conditions, aiming at satisfying the complex aerodynamic heating conditions that arise during reentry, where the lift-drag ratio and drag coefficient are the two objects with the constraints of the stagnation point heat flux and volume efficiency. The NSGA-II is utilized to perform the multi-object configuration optimization. For a given set of configuration parameters generated by the NSGA-II, the three-dimension Navier-Stokes(NS) equation and structure grid Computational Fluid Dynamics(CFD) method are used to compute the lift-drag ratio and drag coefficient, and the Fay-Riddell empirical formula is used to estimate stagnation point heat flux. To improve the computational efficiency, the Kriging surrogate model, with a modified Expected Improvement(EI) sampling strategy to reduce the sample number, is used to replace the CFD computation during the optimization. The optimized results indicate that the Kriging surrogate model can reduce about95% computational cost and the resultant errors are within 7% when compared with the CFD. Thus, the proposed method can be used as a preliminary design tool for reentry capsule configuration optimization.