| 付朋真,张克非,陈江涛,张培红,周桂宇.基于自适应采样的导弹气动数据快速生成技术研究[J].计算力学学报,2025,42(2):243~248 |
|
| 
码上扫一扫！ |
| 基于自适应采样的导弹气动数据快速生成技术研究 |
| Adaptive fast prediction method on aerodynamic data for missile configurations |
| 投稿时间：2023-09-19 修订日期：2024-02-20 |
| DOI：10.7511/jslx20230919001 |
| 中文关键词: 气动数据 代理模型 自适应采样 最大最小缩放距离 CFD数值计算 |
| 英文关键词:aerodynamic data surrogate model adaptive sampling method maximin scaled distance CFD calculations |
| 基金项目:国家数值风洞工程项目(NNW);国家自然科学基金(52175214);国家自然科学基金委员会-中国工程物理研究院NSAF联合项目(U2230208)资助. |
|
| 摘要点击次数: 10 |
| 全文下载次数: 7 |
| 中文摘要: |
| 导弹气动数据是开展导弹气动外形设计的基础,直接使用CFD计算和风洞试验耗时较长,代理模型能够使用现有少量气动数据来预测其他气动数据,采样方法是影响代理模型精度的主要因素。本文基于现有自适应采样方法提出一种基于最大最小缩放距离和交叉验证误差的混合自适应采样方法,针对某型导弹设计,通过少量CFD数值计算结果建立了代理模型,对其他状态下的导弹气动数据进行预测。对比随机采样和交叉验证采样算法,通过多个测试函数和导弹气动数据实例测试,该方法在升力系数指标中误差分别降低35.02%和44.08%,在阻力系数指标中误差分别降低了53.16%和45.17%。 |
| 英文摘要: |
| The aerodynamic design for missiles indispensably relies on the aerodynamic data.Sufficient data is required,and traditionally obtained from CFD calculations and wind-tunnel experiments,which are time-costing in design cycles.A surrogate model can fast produce extra aerodynamic data based on finite existing data,where the accuracy of prediction is mainly limited by the sampling method.In this study,a new hybrid adaptive sampling method based on maximin scaled distance and cross-validation error is proposed and applied to a specific missile.Based on a few CFD data,the extra aerodynamic data is generated by the established surrogate model and compared to the results from the random sampling method and cross-validation sampling method.Results show that this method respectively reduces errors by 35.02% and 44.08% in the lift coefficient index,and 53.16% and 45.17% in the drag coefficient index. |
| 查看全文 查看/发表评论 下载PDF阅读器 |
|
|
|
|
