基于代理模型的水下航行体头型优化设计方法研究

文章采用Kriging代理模型技术针对水下航行体头型优化设计开展研究。采用CFD方法分析了不同头型水下航行体的流体动力性能,将性能参数作为初始样本建立代理模型;根据测试样本点的预测均方误差选择加点策略更新代理模型,提高代理模型预测精度;采用非劣分类遗传算法对阻力和表面压力开展多目标优化,寻找近似最优解。结果表明:基于代理模型技术的水下航行体头型多目标优化设计方法可以有效提高设计效率,获得具有良好水动特性的航行体头型。

Configuration optimization design of underwater vehicle based on surrogate model

Based on Kriging surrogate model, a configuration optimization design method of underwater ve-hicle was studied. A set of different forehead configurations were chosen as original sample points to con-struct the surrogate model. The hydrodynamic forces are calculated by CFD analysis tool. To improve the at-tainable precision, sampling strategy was used according to the MSE (mean square error) of test points. Af-ter that, NSGA-Ⅱ optimization method was implemented under given condition to find the approximate op-timal solution considering minimum drag coefficient and maximum pressure coefficient. The results show that the multi-object optimization design method based on the surrogate model gives a set of Pareto optimal so-lutions. Therefore it can improve the configuration design efficiency of underwater vehicle, and find the ve-hicle configuration with good hydrodynamic performance.